A UAV inspection path planning method for substation equipment

Abstract

This invention discloses a method for planning unmanned aerial vehicle (UAV) inspection paths for substation equipment. This method, which belongs to the field of UAV path planning, includes the following steps: constructing a set of equipment nodes based on the locations of substation equipment; building a UAV path planning model based on a multi-objective optimization algorithm, which is constructed by designing a directional path planning module, an adaptive path variation module, and a non-dominated density sorting module; and inputting the set of equipment nodes into the UAV path planning model based on the multi-objective optimization algorithm for processing to obtain a set of optimized paths that meet the objectives. This invention enables targeted path planning based on the specific substation environment, reduces the consumption of computing resources during the planning process, and ultimately obtains a UAV inspection path for substation equipment.

Description

Technical Field

[0001] The present invention relates to the field of drone path planning, and in particular to a drone inspection path planning method for substation equipment. Background Art

[0002] With the rapid development of drone technology, drones are increasingly being used in various fields. This is particularly true for power equipment inspection, where drone inspections are gradually replacing traditional manual inspections due to their efficiency and convenience. However, current drone inspections often require manual route marking, which is not only time-consuming and labor-intensive, but also inefficient. Therefore, route planning for drone inspections of substation equipment has become an urgent issue.

[0003] When conducting drone inspections in a substation environment, it is usually required to fly along the shortest path to save energy. However, the electromagnetic field generated by the equipment in the substation can easily interfere with the drone's signal. Based on the above practical constraints, the present invention uses the shortest flight path and the smallest communication impact as the drone's inspection goals to improve the inspection efficiency and safety of drones in the complex electromagnetic environment of the substation. The inspection goals are:

[0004] (1) Shortest flight path target : refers to the shortest inspection path length when the drone completes the inspection of the specified equipment;

[0005] (2) Minimum communication impact goal : This means that when a drone is inspecting designated equipment, the amount of electromagnetic interference signals emitted by substation equipment that the drone receives is minimal.

[0006] However, when applying traditional path planning methods to the above problems, there are the following shortcomings:

[0007] (1) Traditional UAV path planning algorithms are inefficient and usually consume a lot of computing resources, resulting in high computational costs;

[0008] (2) Existing optimization algorithms are difficult to adjust according to the distribution of substation equipment, which increases the complexity of performing mutation operations in actual operations;

[0009] (3) Traditional multi-objective optimization algorithms often ignore marginal solutions, which makes it difficult to accurately reflect the distribution of solutions. This may result in some potential high-quality solutions not being obtained, affecting the overall optimization effect.

[0010] To solve the above problems, the present invention proposes a UAV inspection path planning method for substation equipment. Aiming at the two goals of shortest flight path and minimum communication impact, the final inspection path set suitable for UAV inspection of substation equipment is obtained. Summary of the Invention

[0011] This section is intended to summarize some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and in the abstract and title of the present application to avoid obscuring the purpose of this section, the abstract and the title of the invention, and such simplifications or omissions shall not be used to limit the scope of the present invention.

[0012] In view of the above existing problems, the present invention is proposed;

[0013] Therefore, the purpose of the present invention is to provide a UAV inspection path planning method for substation equipment, which can solve the above-mentioned existing problems.

[0014] Explanation of terms:

[0015] Equipment to be inspected: substation equipment that needs to be inspected during the drone inspection process, referred to as equipment.

[0016] Device node and empty node: A cube area divided in the three-dimensional coordinate system space, and the cube area contains the entire device or part of the device, is called a device node; otherwise, the cube area is called an empty node.

[0017] Local path: A node sequence that starts at the drone's takeoff location or a device node and ends at another device node.

[0018] Inspection path: A node sequence composed of local paths and covering all devices.

[0019] Path cost: The cost incurred by a drone when it travels from the starting point to the end point and passes through a certain node on a local path.

[0020] Path mutation: refers to generating a new path by replacing nodes in the inspection path or changing the order between nodes.

[0021] Final inspection path set: the inspection path set selected by the UAV path planning model based on the multi-objective optimization algorithm.

[0022] Highest level points and highest level point sets: In the non-dominated sorting process, the points at the highest level are called highest level points; all points in the highest level form a set called the highest level point set.

[0023] Boundary points and boundary point sets: In the same level other than the highest level, points with extreme ordinates or abscissas are called boundary points; all boundary points constitute a boundary point set.

[0024] Non-boundary points and non-boundary point sets: In any non-top level, points other than boundary points are called non-boundary points; all non-boundary points constitute a non-boundary point set.

[0025] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0026] A method for planning a UAV inspection path for substation equipment includes the following specific steps:

[0027] S1: Collect relevant attribute information of the equipment to be inspected in the substation; map the equipment into a three-dimensional coordinate system; divide the equipment nodes and empty nodes in the space of the three-dimensional coordinate system through rasterization operation, and construct a equipment node set ;

[0028] S2: Construct a UAV path planning model based on a multi-objective optimization algorithm. The construction process of the model is as follows:

[0029] S2.1: Construct the directional path planning module DPP. First, a set of local paths is obtained, and then all the local paths are spliced ​​into an inspection path.

[0030] S2.2: Constructing an adaptive path mutation module APM to perform path mutation on the inspection path to generate multiple new inspection paths, thereby forming a candidate inspection path set;

[0031] S2.3: constructing a non-dominated-density sorting module UDO to sort all candidate inspection paths in the candidate inspection path set, and selecting a final inspection path from the sorting results to form a final inspection path set;

[0032] S3: Set the take-off position of the drone, and for the device node set consisting of all the equipment in a substation, use the drone path planning model based on the multi-objective optimization algorithm to solve the final inspection path set suitable for the drone to inspect the equipment in the substation.

[0033] As a preferred solution of the method for planning a UAV inspection path for substation equipment according to the present invention, the specific steps of S1 are as follows:

[0034] S1.1: Collect the name and number of each device in the substation, measure and record the maximum length, maximum width and maximum height of each device, and record the GPS coordinates of the center point of the device;

[0035] S1.2: Set up a three-dimensional coordinate system with the take-off position of the drone as the origin, map all devices into the three-dimensional coordinate system, and use a cuboid to represent the device, where the length of the cuboid is the maximum length of the device, the width is the maximum width of the device, and the height is the maximum height of the device. The center point coordinates of the device are: The mapping calculation formula is as follows:

[0036]

[0037] in is the GPS coordinate of the drone’s take-off location, is the longitude, is latitude; > is the GPS coordinate of the center point of the device, is the longitude, is latitude; is the maximum height of the device; and is the scaling ratio, and its value is set according to the actual situation;

[0038] S1.3. Divide the three-dimensional coordinate system into multiple equal-sized cubic regions through a gridding operation. Let the origin of the coordinate system be the center of one of the cubic regions. The side length of the cubic region is determined according to the actual situation. If a cubic region contains a part of the device or the entire device, the cube is called a device node; otherwise, it is called an empty node.

[0039] S1.4: Construct a device node set For any device node, if the device node contains the entire device, then add the device node to ; Otherwise, add the device node to After that, nodes containing other parts of the same device will not be added .

[0040] As a preferred solution of the method for planning a UAV inspection path for substation equipment according to the present invention, the specific steps of S2.1 are as follows:

[0041] S2.1.1: Construct a local path set , Each local path in the process ends at a device node; for the local path currently being solved , select a device node from As The end , and will from Remove it, and then solve it as follows :

[0042] S2.1.1.1: If , then starting point Set to the node containing the origin of the 3D coordinate system; otherwise, let Recently acquired and added The local path of The end point as starting point ; Set a candidate node set ,Will join in , and at the same time Set to except The predecessor nodes of all other nodes;

[0043] S2.1.1.2: In In the example, the node with the smallest path cost is selected as the current node ,Will from Remove and add to the collection , It is a collection that stores all nodes that have ever been the current node.

[0044] Each node Path cost The calculation formula is:

[0045]

[0046] in From the starting point Move to Node The cost of movement; It is from arrive The end The heuristic function estimate of ; its calculation formula is:

[0047]

[0048] in yes and The Euclidean distance between is with Related functions for calculating additional costs due to environmental factors; is the weight coefficient.

[0049] S2.1.1.3: and Connect the lines to get the line segments , then make a pass plane and Vertically, for any node , for The node set consisting of all adjacent nodes.

[0050] like The center point and Located on the plane On both sides, from Delete , thus obtaining the node set ;

[0051] For any node ,like Is a device node, then All The adjacent nodes are added to ; Then for any node ,like is a device node, or Already exists in , then from Delete , thus obtaining the node set .

[0052] For any node ,if If any two of the following conditions are met, The preceding node of , and Add to :

[0053] (1) ;

[0054] (2) 、 、 The preceding nodes form an isosceles right triangle, and the hypotenuse of the isosceles right triangle passes through a device node.

[0055] S2.1.1.4: If , then jump to S2.1.1.2; otherwise, As The end point, then Predecessor node As The second to last node of Predecessor node As The third to last node of Add to , and finally Add to local path collection ;

[0056] S2.1.2: All local paths in the

[15] are connected with the device nodes as the connection points to obtain an inspection path.

[0057] As a preferred solution of the UAV inspection path planning method for substation equipment described in the present invention, the specific steps of S2.2 are as follows:

[0058] S2.2.1: For an inspection route , calculate the mutation probability of each node according to the type of each node in the inspection path and sort them, and set a ranking threshold , ranking the mutation probability higher than The nodes are called nodes to be mutated. A path mutation operation is performed on each node to obtain multiple new inspection paths.

[0059] S2.2.2: All the above new inspection routes should be combined with Together they form a set of candidate inspection paths .

[0060] As a preferred solution of the UAV inspection path planning method for substation equipment described in the present invention, the calculation method of the mutation probability is as follows:

[0061] (1) When the node type is an empty node, use the mutation function Counting empty nodes The probability of mutation:

[0062]

[0063] in For For the center of the ball, The number of devices within the sphere with a radius of It is an experience value; is the weight coefficient; is the magnetic field influence coefficient of the equipment; for and The Euclidean distance of is the device node within the sphere;

[0064] (2) When the node type is a device node, a new inspection path is generated by changing the order of the device nodes in the original inspection path.

[0065] The node containing the origin in the three-dimensional coordinate system And all device nodes are projected to plane, and in Plane construction of the first rectangle: select the minimum horizontal coordinate value among all nodes With the maximum vertical coordinate value Construct a coordinate , and select the maximum horizontal coordinate value of all nodes at the same time With the minimum vertical coordinate value Construct the coordinates of another vertex , and let and As the two endpoints of the rectangle's diagonal; according to The position in the rectangle can be divided into two cases for calculating the mutation probability:

[0066] ① Located on the edge of the first rectangle, using the variation function Calculate the mutation probability:

[0067]

[0068] in for The queue composed of all device nodes in, for The number of device nodes in; is the weight coefficient; For the counting function: First, All device nodes in the Sort the distance from small to large, and then compare the sorting results with Match the nodes in and calculate the sorting result and The same number of nodes in the same order;

[0069] ② Otherwise, use the variation function Calculate the mutation probability:

[0070]

[0071] in is the weight coefficient; is the device node; for and The Euclidean distance of .

[0072] As a preferred solution of the UAV inspection path planning method for substation equipment described in the present invention, the process of path variation is as follows:

[0073] (1) When the node to be mutated is an empty node : Get The set of adjacent nodes , for any ,if If one of the adjacent nodes is a device node, then Delete , thus obtaining the node set ; then from Randomly select a node to replace , get a new inspection path;

[0074] (2) When the node to be mutated is a device node: All device nodes in the set ,Will As the input of the DPP module, the DPP module is executed to generate a new inspection path.

[0075] As a preferred solution of the method for planning a UAV inspection path for substation equipment according to the present invention, the specific steps of S2.3 are as follows:

[0076] S2.3.1: In a two-dimensional coordinate system, for any path Use a point To express, is the path length objective function, The communication affects the objective function, and then the non-dominated sorting is used to get the level of each point;

[0077] S2.3.2: Sort all points in the two-dimensional coordinate system from low to high according to hierarchy, wherein within the same hierarchy, sort the points from small to large according to density;

[0078] S2.3.3: Set a ranking threshold , filter out those ranked higher than All points of the inspection paths represented by these points constitute an inspection path set .

[0079] like If there is a non-Pareto optimal solution, each inspection path As the input of the APM module, execute the APM module to obtain a new candidate inspection path set, and then merge all the above new candidate inspection path sets to obtain a candidate inspection path set , jump to Step 2.3.1, and As input to Step 2.3.1;

[0080] otherwise, is the final inspection path set.

[0081] As a preferred solution of the method for planning a UAV inspection path for substation equipment described in the present invention, the and The calculation expression is as follows:

[0082]

[0083]

[0084] in for A device node or empty node in; for The number of nodes in ; for Its adjacent nodes The Euclidean distance between is the magnetic field influence coefficient of the equipment; For Centered on The number of devices within the sphere with a radius of is the experience value; for and The Euclidean distance of The device nodes within the sphere.

[0085] As a preferred solution of the method for planning a UAV inspection path for substation equipment described in the present invention, the density calculation method of the points is as follows:

[0086] (1) If There are multiple points in the level, then

[0087]

[0088] in express density; is the boundary point set; is a non-boundary point set; is the highest level point set;

[0089] when hour, Indicates that at the same level adjacent points; is the weight coefficient; is the area of ​​the second rectangle, and They serve as the two diagonal endpoints of the second rectangle respectively; represents the number of points contained in the second rectangle, where Indicates the area contained in the second rectangle The number of points in the previous level is added by 2 because the second rectangle and These two points are included in the calculation;

[0090] when hour, and Indicates that at the same level adjacent points; is the weight coefficient; is the area of ​​the third rectangle, and They serve as the two diagonal endpoints of the third rectangle; Represents the number of points contained in the third rectangle, where Indicates the third rectangle contains The number of points in the previous level is added 3 because the third rectangle 、 and These three points are included in the calculation;

[0091] when hour: The value of .

[0092] (2) If There is only one point in the level, then The density of has no effect on the sorting, so .

[0093] Compared with existing technologies:

[0094] (1) The present invention designs a directional path planning module, which reduces the computational complexity of the path planning process and improves the efficiency of path planning;

[0095] (2) The present invention designs an adaptive path variation module that can dynamically adjust the variation strategy according to the actual situation of the nodes in the substation and perform targeted path variation, thereby generating inspection paths with a higher degree of diversity;

[0096] (3) The present invention designs a density calculation method, which reflects the specific distribution of the inspection path in the target optimization process through the density of each point, and at the same time realizes the effective processing of the inspection path distribution in the edge area, ensuring the comprehensiveness of the distribution information in space. BRIEF DESCRIPTION OF THE DRAWINGS

[0097] Figure 1 It is a schematic diagram of the process of the present invention.

[0098] Figure 2 This is a schematic diagram of the execution flow of the drone path planning model of the present invention.

[0099] Figure 3 Schematic diagram of the local path set construction process of the present invention.

[0100] Figure 4 Schematic diagram of adjacent node screening according to the present invention.

[0101] Figure 5 The schematic diagram of the front node is changed for the present invention.

[0102] Figure 6 This is a schematic diagram of the execution flow of the APM module of the present invention.

[0103] Figure 7 This is a schematic diagram of the distribution of device nodes in the present invention.

[0104] Figure 8 This is a second schematic diagram of the device node distribution of the present invention.

[0105] Figure 9 For the present invention Schematic diagram of density calculation.

[0106] Figure 10 For the present invention Schematic diagram of density calculation. DETAILED DESCRIPTION

[0107] To make the above-mentioned objects, features, and advantages of the present invention more clearly understood, the following detailed description of the specific embodiments of the present invention is given in conjunction with the accompanying drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in this field without creative work should fall within the scope of protection of the present invention. Example

[0108] See also Figures 1-10 , which is the first embodiment of the present invention, provides a method for planning a UAV inspection path for substation equipment, comprising:

[0109] When conducting drone inspections in substation environments, drones are typically required to fly along the shortest possible path to save energy. However, the electromagnetic fields generated by substation equipment can easily interfere with drone signals.

[0110] This application provides a method that can effectively solve the above-mentioned problems. Next, we will combine multiple embodiments to elaborate on how to implement a drone inspection path planning method for substation equipment.

[0111] Figure 1 A flowchart of a method for planning a UAV inspection path for substation equipment is shown, including:

[0112] S1: Constructing a device node set for the substation: Collecting the relevant attribute information of the equipment to be inspected (hereinafter referred to as equipment) in the substation: name, number, location, maximum length, maximum width and maximum height; setting a three-dimensional coordinate system, mapping the location of the equipment to the three-dimensional coordinate system, dividing the device nodes and empty nodes in the space of the three-dimensional coordinate system through rasterization operation, and constructing a device node set ;

[0113] In the embodiment of the present application, the process of constructing the device node set of the substation is as follows:

[0114] S1.1: Collect the name and number of each device in the substation, measure and record the maximum length, maximum width and maximum height of each device, and record the GPS coordinates of the center point of the device;

[0115] S1.2: Set up a three-dimensional coordinate system with the take-off position of the drone as the origin of the coordinate system, map all devices into the three-dimensional coordinate system, and abstractly represent the device as a cuboid, where the length of the cuboid is the maximum length of the device, the width is the maximum width of the device, and the height is the maximum height of the device. The center point coordinates of the device are The mapping calculation formula is as follows:

[0116]

[0117] in is the location attribute of the device; is the GPS coordinate of the drone’s take-off location, is the longitude, is latitude; > is the GPS coordinates of the device, is the longitude, is latitude; is the maximum height of the device; and is the scaling ratio, and its value is set according to the actual situation.

[0118] For example, assuming the GPS coordinates of the drone's takeoff location are <22.837398,113.557785>, a device The GPS coordinates are <22.845229,113.561589>, The maximum height is 3 meters, Set to 0.001, Set it to 2, and get The center point coordinates in the three-dimensional coordinate system< > = <-7.831, -3.804, 1.5>.

[0119] S1.3 divides the three-dimensional coordinate system space into multiple equal-sized cubic regions through a gridding operation, with the origin of the coordinate system being the center of one of the cubic regions. The side lengths of the cubic regions are determined based on actual conditions. If a cubic region contains a part of a device or the entire device, the cube is called a device node; otherwise, it is called an empty node.

[0120] For example, suppose the coordinates of a cube area are <-7, -3, 1> ~ <-9, -5, 3>, The maximum length is 2 meters, the maximum width is 2 meters, and the maximum height is 3 meters. Represented as a cuboid, according to The center point coordinates of > = <-7.831, -3.804, 1.5>, the coordinates of the above rectangular parallelepiped are <-6.831, -2.804, 0> ~ <-8.831, -4.804, 3>. Among them, the coordinates of the above rectangular parallelepiped <-7, -3, 1> ~ <-8.831, -4.804, 3> are within the above cube area, so the above cube area is The device node.

[0121] S1.4: Construct a device node set For any device node, if the device node contains the entire device, then add the device node to ; Otherwise, add the device node to After that, nodes containing other parts of the same device will not be added .

[0122] S2: Construct a UAV path planning model based on a multi-objective optimization algorithm.

[0123] In the embodiment of the present application, in order to realize the multi-objective inspection path planning of the UAV in the substation, a UAV path planning model based on the multi-objective optimization algorithm is constructed. The execution process of the model is as follows: Figure 2 The model obtains multiple inspection paths through path planning and and Construct the corresponding objective function, optimize the inspection path under the constraints of the above two objective functions, and finally obtain a final inspection path set.

[0124] The UAV path planning model consists of three modules: a directional path planning module, an adaptive path variation module, and a non-dominated-density sorting module. The directional path planning module is used to determine an inspection path; the adaptive path variation module is used to construct a set of candidate inspection paths; and the non-dominated-density sorting module is used to sort the paths and select the final inspection path set.

[0125] S2.1: Build a directional path planning module to plan inspection paths:

[0126] Use a directional path planning module DPP (Directional Path Planning) to As the input of DPP, starting from the starting point, the adjacent nodes are screened according to the center position of the adjacent nodes, and the screened adjacent nodes are gradually explored until the end position is explored. Multiple local paths are planned, and then all the local paths are spliced ​​to obtain an inspection path.

[0127] The construction process of the DPP module is as follows:

[0128] by As input, a set of local paths is first solved, and then all the local paths in the above set are spliced ​​together to obtain an inspection path as output.

[0129] S2.1.1: Construct a local path set , Each local path in the process ends at a device node; for the local path currently being solved, , its construction process is as follows Figure 3 As shown, from Select a device node ,Will Set to The end , and from Remove it, and then solve it as follows :

[0130] S2.1.1.1: If , then starting point is set to the node containing the origin of the 3D coordinate system; otherwise, There is at least one local path in Indicates that it has been recently obtained and added The local path of The end point as starting point . Set a candidate node set ,Will join in At the same time, in the three-dimensional coordinate system, Set to except The predecessor nodes of all nodes except ;

[0131] S2.1.1.2: Apply cost functions calculate The path cost of each node in Select the node with the smallest path cost as the current node ,Will from Remove and add to the collection , It is a collection that stores all nodes that have ever been the current node.

[0132] Each node Path cost The calculation formula is:

[0133]

[0134] in From the starting point Move to Node The cost of movement; It is from arrive The end The heuristic function estimate of ; its calculation formula is:

[0135]

[0136] in yes and The Euclidean distance between is with Related functions for calculating additional costs due to environmental factors; is the weight coefficient.

[0137] The cost function is used to evaluate the total cost of each node in order to find the optimal path from the starting point to the end point.

[0138] A heuristic function is a function used to estimate the shortest path cost from the current node to the destination. It guides the search process, enabling the algorithm to prioritize nodes that are more likely to be close to the destination, thereby improving search efficiency.

[0139] Euclidean distance is a geometric distance used to measure the straight-line distance between two points, representing the shortest path length between the two points. In three-dimensional space, its formula is:

[0140]

[0141] S2.1.1.3 will All adjacent nodes constitute a node set ,in for An adjacent node of for The number of adjacent nodes;

[0142] Will and Connect the lines to get a line segment , and make a pass The plane and Vertical. For any node ,like The center point of Located on both sides of the plane, Delete , thus obtaining the node set .

[0143] For any node ,like For device nodes, then Find and All adjacent nodes are added to Then Filter: For any node ,when is a device node, or Already in In Delete .Will After the above processing, the node set is obtained .

[0144] For example, Figure 4 Shown is a top plan view of a three-dimensional coordinate system. It's the end point. is the current node, and Is the device node. and Connect the lines to get the line segments , and make a plane and vertical, All adjacent nodes constitute a node set .node 、 、 and Located on the plane on one side, and lie in The other side of Delete, 、 、 and (exist Figure 4 In the example, the nodes with dotted lines represent the deleted nodes). After the above processing, the node set is obtained .

[0145] because and Are all device nodes, then Find and 、 Adjacent nodes 、 ,Will and join in middle( Already exists in So we only need to Add to ), from which we get Then Filter: Because and Are all device nodes, then from Delete and , thus we get .

[0146] For any node ,like If any of the following conditions are met, The preceding node of , and Add to :

[0147] (1) ;

[0148] (2) 、 、 The preceding nodes form an isosceles right triangle, and the hypotenuse of the isosceles right triangle passes through a device node.

[0149] The above changes The purpose of the preceding node operation is to obtain a local path based on the preceding node in the next step; at the same time, it reduces the risk of the drone colliding with the equipment during flight.

[0150] For example, Figure 5 The figure shows a top view of the three-dimensional coordinate system. As the current node, yes The adjacent nodes of yes The predecessor node of and is any device node. Fly to , then it is possible that Scrape, so need to change The predecessor node of . The three form an isosceles right triangle, and the hypotenuse After the device node , then The preceding node of . The drone from Fly to It will not pass through any device nodes, reducing the risk of collision between drones.

[0151] S2.1.1.4: If , indicating that the end of the local path has not been explored, jump to S2.1.1.2; otherwise, As The end point, Predecessor node As The second to last node of Predecessor node As The third to last node of Add to Finally, we get The node sequence is { },Will Add to local path collection ;

[0152] S2.1.2: All local paths in the inspection path are spliced ​​according to the starting and ending points of the paths to obtain an inspection path.

[0153] For example, assuming ,in The starting point is the node containing the origin of the coordinate system. for The end point of It is also The starting point. As the connection point, and Splice to get a longer local path Repeat the above steps to and After splicing, an inspection path is obtained.

[0154] S2.2: Build an adaptive path variation module to create a set of candidate inspection paths:

[0155] Use an adaptive path mutation module APM (Adaptive Path Mutation) to As input, by designing and applying the mutation function, we get the nodes with the highest mutation function value, and perform path mutation according to the node type to generate multiple new inspection paths, and finally generate a set of candidate inspection paths.

[0156] The construction process of the APM module is as follows:

[0157] by As input, the inspection path is then mutated to obtain a set of candidate inspection paths as output.

[0158] Mutation, in evolutionary algorithms, involves randomly changing certain genes within individuals to introduce diversity and explore new solution spaces. This process helps prevent the algorithm from becoming trapped in local optima and enhances the population's ability to search for the global optimal solution. It should be noted that in this module, individuals are equivalent to paths, and genes are equivalent to nodes within those paths.

[0159] Figure 6 The execution flow chart of the APM module is shown as follows:

[0160] S2.2.1: For , according to the type of nodes in the inspection path, the corresponding mutation function is used to calculate the mutation probability of each node. Then the nodes are sorted according to the mutation probability and a ranking threshold is set. , ranking the mutation probability higher than The nodes are called nodes to be mutated. Perform path mutation to obtain multiple new inspection paths.

[0161] Among them, the corresponding mutation function is designed according to different node types:

[0162] (1) When the node type is an empty node, use the mutation function To calculate the mutation probability of the node. The design is as follows:

[0163]

[0164] in Represents an empty node; For For the center of the ball, The number of devices within the sphere with a radius of It is an experience value; is the weight coefficient; is the magnetic field influence coefficient of a certain device; for and The Euclidean distance of is a device node within the sphere;

[0165] (2) When the node type is a device node, since the device node is a node that must be passed through in the inspection path, a new inspection path cannot be generated by replacing the device node. Instead, a new inspection path is generated by changing the order of the device nodes in the original inspection path. Therefore, the empty nodes in the inspection path are eliminated, and a sequence consisting of device nodes is obtained, which is called a device node queue. At this time, the mutation function will be designed for the device node queue.

[0166] The node containing the origin in the three-dimensional coordinate system And all device nodes are projected to plane, and in Plane construction of the first rectangle: select the minimum horizontal coordinate value among all nodes With the maximum vertical coordinate value Construct a vertex coordinate , and select the maximum horizontal coordinate value of all nodes at the same time With the minimum vertical coordinate value Construct the coordinates of another vertex , and let and As the two endpoints of the rectangle's diagonal; according to The position in the rectangle can be divided into two cases for calculating the mutation probability:

[0167] ①、 Located on the edge of the first rectangle (such as Figure 7 As shown, the nodes drawn with dotted lines are , the nodes drawn with solid lines are device nodes). At this time, the mutation function is used To calculate the mutation probability of the device node queue:

[0168]

[0169] in for The device node queue; for The number of device nodes contained in; is the weight coefficient; is the counting function, Each device node in Sort the straight-line distances from smallest to largest, and then compare the sorting results with Matching is performed, and the sorting result is calculated and The same number of nodes in the same order, that is, when the sorting result is the same as When the nodes in the same order are the same, the counting function value is increased by 1.

[0170] For example, assuming , 、 、 、 The straight-line distances from the drone's takeoff position are 3, 2, 1, and 5. The results are sorted from smallest to largest: ,Will and The result is 2, because the nodes in the second order position in both queues are , the nodes in the fourth order position are , while other nodes in the same order are different. Weight coefficient Set to 4, then the mutation probability is = .

[0171] ②、 Not on the edge of the first rectangle (such as Figure 8 As shown, the nodes drawn with dotted lines are , the nodes drawn with solid lines are device nodes), at this time, the mutation function is used To calculate the mutation probability of the device node queue:

[0172]

[0173] in, for The device node queue; for The number of device nodes contained in; is the weight coefficient; for Any device node in; for and The Euclidean distance of for The sum of the distances to each device node; for The average distance to each device node.

[0174] Different methods should be used to mutate different types of nodes:

[0175] (1) When the node to be mutated is an empty node : Get The set of adjacent nodes , for any ,if If one of the adjacent nodes is a device node, then Delete , thus obtaining the node set ; then from Randomly select a node to replace , get a new inspection path;

[0176] (2) When the node to be mutated is a device node: All device nodes in the set ,Will As the input of the DPP module, the DPP module is executed to generate a new inspection path.

[0177] S2.2.2: Compare the new inspection path obtained above with Together they form a set of candidate inspection paths ,in For a certain path, Indicates the number of paths.

[0178] S2.3: Construct a non-dominated-density sorting module to select the final inspection path set:

[0179] The present invention designs a non-dominated-density ordering module UDO (Undominated-density ordering) to screen out the final inspection path set. As input, the inspection paths are sorted by using non-dominated sorting and density calculation, and finally multiple inspection paths within a certain ranking threshold are screened out, and the above inspection paths are merged into a set as the final inspection path set.

[0180] S2.3.1: Construct a two-dimensional coordinate system for any path ,Will Placed in a two-dimensional coordinate system, using the path length objective function ,get As The horizontal axis of the communication influence objective function ,get As A point in the two-dimensional coordinate system represents an inspection path. , and use non-dominated sorting to get the level of each point.

[0181] Nondominated sorting is a method used to classify solutions in multi-objective optimization. Its goal is to rank all paths (i.e., the population) based on their nondomination relationships. First, the nondominated counts of all paths in the population are calculated. Then, paths with the same nondominated counts are grouped together into a hierarchy. Finally, the hierarchies are sorted in ascending order of nondominated counts to determine the hierarchy's size. The nondominated count refers to the total number of paths dominated by other paths; being dominated by other paths means that any objective function value of a path is inferior to that of other paths.

[0182] The calculation formula is designed as follows;

[0183]

[0184] in For any inspection path; for A device node or empty node in; for The number of nodes in ; for Its adjacent path nodes The Euclidean distance between express The total length of

[0185] The calculation formula is designed as follows:

[0186]

[0187] in, For any inspection path; for A device node or empty node in; for The number of nodes in ; is the magnetic field influence coefficient of a certain device; For Centered on The number of devices within the sphere with a radius of is the experience value; for and The Euclidean distance of A device node within the sphere.

[0188] S2.3.2: Sort all points from low to high according to the hierarchy. Within the same hierarchy, sort the points from small to large according to their density. At this time, the density of the points will be calculated using the density formula Perform calculations.

[0189] (1) If There are multiple points in the level, then The calculation formula is as follows:

[0190]

[0191] in, for Any point in is the boundary point set; is a non-boundary point set; is the highest level point set;

[0192] ①, when When, such as Figure 9 As shown, Indicates that at the same level adjacent points; is the weight coefficient; represents the area of ​​the second rectangle, where and As the two endpoints of the diagonal of the second rectangle; Represents the number of points contained in the second rectangle, where Indicates the area contained in the second rectangle The number of points in the previous level is added by 2 because the second rectangle and These two points are included in the calculation. is a boundary point, so the calculated density can reflect the distribution of paths in the edge area;

[0193] ② When When, such as Figure 10 As shown, and Indicates that at the same level adjacent points; is the weight coefficient; represents the area of ​​the third rectangle, where and As the two endpoints of the diagonal of the third rectangle; Represents the number of points contained in the third rectangle, where Indicates the third rectangle contains The number of points in the previous level is added 3 because the third rectangle 、 and These three points are included in the calculation. At this time, due to Since it is a non-boundary point, the calculated density can reflect the distribution of paths in the local area;

[0194] ③. When When: The value is Because when When it is the highest level point, and are higher than other points, so they need to be eliminated in the subsequent screening process. .Will The density of , to ensure that it is placed last in the sorting results, so as to achieve Optimization.

[0195] (2) If There is only one point in the level, then The density of has no effect on the sorting, so .

[0196] S2.3.3: Set a ranking threshold , filter out those ranked higher than All points of the inspection paths represented by these points constitute an inspection path set .

[0197] like If there is a non-Pareto optimal solution, each inspection path As the input of the APM module, execute the APM module to obtain a new candidate inspection path set, and then merge all the above new candidate inspection path sets to obtain a new candidate inspection path set , jump to Step 2.3.1, and As input to Step 2.3.1;

[0198] otherwise, is the final inspection path set.

[0199] Pareto optimality is a key concept in multi-objective optimization. In multi-objective optimization problems, there are often multiple conflicting objectives, meaning that improving one objective may degrade another. A Pareto optimal solution is one that cannot be further improved without compromising any of the objectives. These Pareto optimal solutions form a Pareto frontier, the set of points in the objective space that represents the optimal trade-off. In this module, Pareto optimal solutions are defined as any point in the lowest level (the first level).

[0200] S3: Set the take-off position of the drone, and use the drone path planning model based on the multi-objective optimization algorithm to solve the device node set composed of all devices in a substation to obtain the final inspection path set.

[0201] Based on the above, the present invention designs a DPP module to generate an inspection path, thereby solving the problem of high computational resource consumption of traditional drone path planning algorithms. In view of the distribution of equipment within the substation, the present invention further designs an APM module to generate multiple inspection paths through path variation, solving the problem of insufficient inspection paths. Finally, the present invention utilizes the UDO module to sort all inspection paths and obtain the final optimized inspection path set. In this process, the design of the density function is used to analyze the actual distribution of each solution, while also solving the problem of marginal solutions being ignored, resulting in the inability to obtain potential high-quality paths.

[0202] Although the present invention has been described above with reference to embodiments, various modifications may be made thereto and equivalent components may be substituted without departing from the scope of the present invention. In particular, as long as there are no structural conflicts, the various features of the embodiments disclosed herein may be combined with each other in any manner, and the omission of an exhaustive description of such combinations in this specification is solely for the sake of space and resource conservation. Therefore, the present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A UAV inspection path planning method for substation equipment, characterized in that: The specific steps are as follows: S1: Collect relevant attribute information of the equipment to be inspected in the substation; map the equipment into a three-dimensional coordinate system; divide the equipment nodes and empty nodes in the space of the three-dimensional coordinate system through rasterization operation, and construct a equipment node set ; S2: Construct a UAV path planning model based on a multi-objective optimization algorithm. The construction process of the model is as follows: S2.1: Construct the directional path planning module DPP. First, a set of local paths is obtained, and then all the local paths are spliced ​​into an inspection path. S2.2: Constructing an adaptive path mutation module APM to perform path mutation on the inspection path to generate multiple new inspection paths, thereby forming a candidate inspection path set; S2.3: constructing a non-dominated-density sorting module UDO to sort all candidate inspection paths in the candidate inspection path set, and selecting a final inspection path from the sorting results to form a final inspection path set; S3: Set the take-off position of the drone, and for the device node set consisting of all the equipment in a substation, use the drone path planning model based on the multi-objective optimization algorithm to solve the final inspection path set suitable for the drone to inspect the equipment in the substation.

2. A method for planning a UAV inspection path for substation equipment according to claim 1, characterized in that: The specific steps of S1 are as follows: S1.1: Collect the name and number of each device in the substation, measure and record the maximum length, maximum width and maximum height of each device, and record the GPS coordinates of the center point of the device; S1.2: Set up a three-dimensional coordinate system with the take-off position of the drone as the origin, map all devices into the three-dimensional coordinate system, and use a cuboid to represent the device, where the length of the cuboid is the maximum length of the device, the width is the maximum width of the device, and the height is the maximum height of the device. The center point coordinates of the device are: The mapping calculation formula is as follows: in is the GPS coordinate of the drone’s take-off location, is the longitude, is latitude; > is the GPS coordinate of the center point of the device, is the longitude, is latitude; is the maximum height of the device; and is the scaling ratio, and its value is set according to the actual situation; S1.

3. Divide the three-dimensional coordinate system into multiple equal-sized cubic regions through a gridding operation. Let the origin of the coordinate system be the center of one of the cubic regions. The side length of the cubic region is determined according to the actual situation. If a cubic region contains a part of the device or the entire device, the cube is called a device node; otherwise, it is called an empty node. S1.4: Construct a device node set For any device node, if the device node contains the entire device, then add the device node to ; Otherwise, add the device node to After that, nodes containing other parts of the same device will not be added .

3. The method for planning a UAV inspection path for substation equipment according to claim 1, characterized in that: The construction process of the S2.1 directional path planning module DPP is as follows: S2.1.1: Construct a local path set , Each local path in the process ends at a device node; for the local path currently being solved , select a device node from As The end , and will from Remove it, and then solve it as follows : S2.1.1.1: If , then starting point Set to the node containing the origin of the 3D coordinate system; otherwise, let Recently acquired and added The local path of The end point as starting point ; Set a candidate node set ,Will join in , and at the same time Set to except The predecessor nodes of all other nodes; S2.1.1.2: In In the example, the node with the smallest path cost is selected as the current node ,Will from Remove and add to the collection , It is a collection that stores all nodes that have ever been the current node; Each node Path cost The calculation formula is: in From the starting point Move to Node The cost of movement; It is from arrive The end The heuristic function estimate of ; its calculation formula is: in yes and The Euclidean distance between is with Related functions for calculating additional costs due to environmental factors; is the weight coefficient; S2.1.1.3: and Connect the lines to get the line segments , then make a pass plane and Vertically, for any node , for The node set consisting of all adjacent nodes; like The center point and Located on the plane On both sides, from Delete , thus obtaining the node set ; For any node ,like Is a device node, then All The adjacent nodes are added to ; Then for any node ,like is a device node, or Already exists in , then from Delete , thus obtaining the node set ; For any node ,if If any two of the following conditions are met, The preceding node of , and Add to : （1） ； (2) 、 、 The preceding nodes form an isosceles right triangle, and the hypotenuse of the isosceles right triangle passes through a device node; S2.1.1.4: If , then jump to S2.1.1.2; otherwise, As The end point, then Predecessor node As The second to last node of Predecessor node As The third to last node of Add to , and finally Add to local path collection ; S2.1.2: All local paths in the [15] are connected with the device nodes as the connection points to obtain an inspection path.

4. The method for planning a UAV inspection path for substation equipment according to claim 1, characterized in that: The specific steps of constructing the adaptive path variation module APM to perform path variation in S2.2 are as follows: S2.2.1: For an inspection route , calculate the mutation probability of each node according to the type of each node in the inspection path and sort them, and set a ranking threshold , ranking the mutation probability higher than The nodes are called nodes to be mutated. A path mutation operation is performed on each node to obtain multiple new inspection paths. S2.2.2: All the above new inspection routes should be combined with Together they form a set of candidate inspection paths .

5. The construction process of the adaptive path variation module APM according to claim 4 is characterized in that: The calculation method of the mutation probability is as follows: (1) When the node type is an empty node, use the mutation function Counting empty nodes The probability of mutation: in For For the center of the ball, The number of devices within the sphere with a radius of It is an experience value; is the weight coefficient; is the magnetic field influence coefficient of the equipment; for and The Euclidean distance of is the device node within the sphere; (2) When the node type is a device node, a new inspection path is generated by changing the order of the device nodes in the original inspection path: The node containing the origin in the three-dimensional coordinate system And all device nodes are projected to plane, and in Plane construction of the first rectangle: select the minimum horizontal coordinate value among all nodes With the maximum vertical coordinate value Construct a coordinate , and select the maximum horizontal coordinate value of all nodes at the same time With the minimum vertical coordinate value Construct the coordinates of another vertex , and let and As the two endpoints of the rectangle's diagonal; according to The position in the rectangle can be divided into two cases for calculating the mutation probability: ① Located on the edge of the first rectangle, using the variation function Calculate the mutation probability: in for The queue composed of all device nodes in, for The number of device nodes in; is the weight coefficient; For the counting function: First, All device nodes in the Sort the distance from small to large, and then compare the sorting results with Match the nodes in and calculate the sorting result and The same number of nodes in the same order; ② Otherwise, use the variation function Calculate the mutation probability: in is the weight coefficient; is the device node; for and The Euclidean distance of .

6. The process of constructing the adaptive path variation module APM according to claim 4 is characterized in that: The process of the path mutation operation is as follows: (1) When the node to be mutated is an empty node : Get The set of adjacent nodes , for any ,if If one of the adjacent nodes is a device node, then Delete , thus obtaining the node set ; then from Randomly select a node to replace , get a new inspection path; (2) When the node to be mutated is a device node: All device nodes in the set ,Will As the input of the DPP module, the DPP module is executed to generate a new inspection path.

7. The method for planning a UAV inspection path for substation equipment according to claim 1, characterized in that: The specific steps of constructing the non-dominated-density ordering module UDO in S2.3 and screening the final inspection path set are as follows: S2.3.1: In a two-dimensional coordinate system, for any path Use a point To express, is the path length objective function, The communication affects the objective function, and then the non-dominated sorting is used to get the level of each point; S2.3.2: Sort all points in the two-dimensional coordinate system from low to high according to hierarchy, wherein within the same hierarchy, sort the points from small to large according to density; S2.3.3: Set a ranking threshold , filter out those ranked higher than All points of the inspection paths represented by these points constitute an inspection path set ; like If there is a non-Pareto optimal solution, each inspection path As the input of the APM module, execute the APM module to obtain a new candidate inspection path set, and then merge all the above new candidate inspection path sets to obtain a candidate inspection path set , jump to Step 2.3.1, and As input to Step 2.3.1; otherwise, is the final inspection path set.

8. The non-dominated density ordering module (UDO) according to claim 7, characterized in that: described and The calculation expression is as follows: in for A device node or empty node in; for The number of nodes in ; for Its adjacent nodes The Euclidean distance between is the magnetic field influence coefficient of the equipment; For Centered on The number of devices within the sphere with a radius of is the experience value; for and The Euclidean distance of The device nodes within the sphere.

9. The non-dominated density ordering module (UDO) according to claim 7, characterized in that: The density of the points is calculated as follows: (1) If There are multiple points in the level, then in express density; is the boundary point set; is a non-boundary point set; is the highest level point set; when hour, Indicates that at the same level adjacent points; is the weight coefficient; is the area of ​​the second rectangle, and They serve as the two diagonal endpoints of the second rectangle respectively; represents the number of points contained in the second rectangle, where Indicates the area contained in the second rectangle The number of previous level points; when hour, and Indicates that at the same level adjacent points; is the weight coefficient; is the area of ​​the third rectangle, and They serve as the two diagonal endpoints of the third rectangle; Represents the number of points contained in the third rectangle, where Indicates the third rectangle contains The number of previous level points; when hour: The value of ; (2) If There is only one point in the level, then The density of has no effect on the sorting, so .

Images