Method and system for planning dynamic route of unmanned aerial vehicle of transformer substation

Abstract

The invention relates to the technical field of unmanned aerial vehicle inspection, in particular to a transformer substation unmanned aerial vehicle dynamic route planning method and system. A patrol demand is converted into a detailed point location data set, a unified safety patrol point location set is generated by using coordinate conversion, safety radius calculation and shooting parameter setting, safety constraints are introduced through a path selection probability formula based on an improved ant colony algorithm, and a multi-ant colony parallel search and an adaptive pheromone updating mechanism are combined to obtain a safety patrol point location set. The method comprises the steps of generating an optimized shortest route, then performing to-be-examined identification generation and verification on an optimized waypoint set, performing path smoothing processing on the waypoint set, fitting a starting point, an end point and a middle patrol point through a third-order Bezier curve, smoothly joining the route, meeting a safe steering angle, path smoothness and a safe radius constraint, and finally obtaining the optimal route. According to the method, the requirements of the substation patrol task on safety, high efficiency and accuracy are met.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) inspection, and particularly to a method and system for dynamically planning UAV flight routes in a substation. Background Art

[0002] With the development of the intellectualization and automation of the power system, as a key node in power transmission, the inspection and maintenance of substations are crucial for the safe operation of the power grid. Traditional manual inspection methods have problems such as low efficiency, high danger, and incomplete data collection, and are difficult to meet the requirements of the efficient and stable operation of modern power grids. Due to its advantages such as high flexibility, wide coverage, and accurate data collection, UAV technology has gradually become an important means for substation inspection.

[0003] A Chinese invention application with the publication number CN116594426A discloses a method and system for planning UAV inspection routes in a substation. The method includes: obtaining three-dimensional point cloud data of the substation within the inspection area, and dividing the substation into several sub-areas in the three-dimensional point cloud data according to equipment intervals; extracting inspection points within each sub-area in the three-dimensional point cloud data, and generating corresponding waypoints based on the inspection points; planning intra-area routes and intra-area safety channels for each sub-area according to the waypoints, and planning cross-area safety channels between each sub-area according to the intra-area safety channels of all sub-areas; generating inspection route model data according to the cross-area safety channels, the intra-area safety channels within each sub-area, and the intra-area routes; generating dynamic inspection routes according to the inspection route model data and the inspection plan of the substation. The inspection method of the present invention is flexible and efficient.

[0004] However, in actual production and life, when a UAV flies in a substation, it generally can only fly according to a preset route. However, the route planning is relatively rigid, unable to flexibly adjust the observation target and most likely save the flight time. In addition, during the execution of complex tasks, the UAV also needs to meet multiple constraint conditions such as safety radius, path smoothness, and flight altitude stability to avoid flight deviation or collision accidents caused by unreasonable paths. Therefore, it is of great research value and practical application significance to propose an efficient and dynamic UAV route planning method to meet the safety and efficiency requirements of substation equipment inspection. Summary of the Invention

[0005] The object of the present invention is to propose a method and system for dynamically planning UAV flight routes in a substation for the problems existing in the background art.

[0006] The technical solution of the present invention: A method for dynamically planning UAV flight routes in a substation includes the following specific implementation steps:

[0007] S1. Convert the inspection requirements into a detailed set of point data, and through steps such as coordinate transformation, safety radius calculation, and shooting parameter setting, generate an output set of safe inspection points that meet the inspection requirements;

[0008] S2. Combine heuristic information with dynamic constraint conditions, and through path selection probability, dynamic safety constraints, and multi-ant colony parallel search mechanism, complete the optimization and streamlining of the inspection point path, output the optimized path point set, and generate a to-be-reviewed identifier for the optimized path point set;

[0009] S3. Based on the to-be-reviewed identifier, review the overall integrity of the received path point set. Based on the optimized path point set, use the third-order Bezier curve for path smoothing fitting to ensure smooth transitions between the starting point, ending point, and intermediate points. Consider flight constraints during the path smoothing process, including the maximum turning angle, path smoothness, and safety radius. Insert safety connection points and intermediate points to ensure that the interval between every two path points meets the safety radius requirement, and adjust the turning angle to meet the maximum safe turning angle constraint, finally generating the inspection path points;

[0010] S4. Use the density-based clustering algorithm to perform spatial clustering on the path points, divide the path points into multiple clusters, and for each cluster, use an improved TSP algorithm to optimize the path order to minimize the total path length. The path selection is dynamically adjusted according to the pheromone concentration, heuristic factor, path length, and safety constraint factor. Through clustering and path optimization, obtain the global temporary optimal flight path;

[0011] S5. Upload the global temporary optimal flight path to the UAV control system, and the UAV flies sequentially according to the flight path, visits all inspection target points, and performs image acquisition and task execution at each target point;

[0012] S6. Use the collected images and data for image processing, anomaly detection, and generate an analysis report, output the recognition results, and visualize the planned flight path, the collected images, and the recognition results accordingly.

[0013] Preferably, the generation process of the optimized path point set is as follows:

[0014] S21. Input data: Set of safe inspection points: P = {(x′ i , y′ i , z i , δ i , θ i , d i , R i )|i = 1, 2,..., n};

[0015] Define the starting safety point S = (x′0, y′0, z0);

[0016] Define the termination safety point E = (x′ n , y′ n , z n );

[0017] Among them, n represents the number of safety inspection points;

[0018] S22. Objective: Plan the shortest route R passing through all inspection points, minimize the path length to the greatest extent while ensuring path safety, and output the sequence of route points and the total path length after refined optimization;

[0019] S23. Dynamic path constraint optimization, introduce a safety radius constraint in the path selection process to prevent the UAV from approaching obstacles or entering dangerous areas, and the path selection probability P ij The formula is:

[0020]

[0021] In the formula, τ ij represents the pheromone concentration of the path i→j; η ij represents the path heuristic factor; d ij represents the path length; ψ ij represents the safety constraint factor. If the path i→j does not satisfy the safety radius constraint, then ψ ij <1, otherwise it is 1; α and β represent the parameters controlling the weights of pheromone and heuristic factor; U represents the set of remaining points that the ant can currently access;

[0022] S24. Multi-ant colony parallel search mechanism: Introduce multiple ant colonies to execute search tasks in parallel, increase the global search ability of the algorithm, effectively avoid the problem of local optimal solutions, each ant colony independently searches for paths, and finally selects the global optimal path for output;

[0023] S25. Adaptive pheromone update mechanism, dynamically adjust the pheromone update amount according to the path quality, strengthen the better paths and weaken the inferior paths:

[0024] τ ij ←(1 - ρ)·τ ij +ρ·Δτ ij ;

[0025]

[0026] In the formula, ρ represents the pheromone evaporation coefficient; Q represents the pheromone intensity; L represents the path length; f adj represents the dynamic adjustment factor, which is adaptively adjusted based on the current iteration round and path quality;

[0027] S26. Output the optimized path point set based on the improved ant colony algorithm.

[0028] Preferably, the process of outputting the optimized path point set based on the improved ant colony algorithm is as follows:

[0029] S31. Initialization: Input the inspection point data P, initialize the number of ants m, the maximum number of iterations T, the pheromone concentration τ ij , the heuristic factor η ij and set the safety radius constraint δ i ;

[0030] S32. Ant path search:

[0031] Each ant starts from the starting point S and visits the inspection points in sequence based on the path selection probability P ij according to the formula:

[0032] S3201. Calculate the selection probability P of each optional path ij , and select the next inspection point to visit based on this probability;

[0033] S3202. Selection strategy: The ant selects the next point according to the probability P of the current path ij . If there are multiple selectable paths, it is more inclined to select the path with a higher pheromone concentration;

[0034] S3203. Safety constraint: The selected path d ij needs to meet the minimum safety radius δ i requirement to ensure the flight safety of the drone;

[0035] Path execution and recording: Each ant passes through multiple inspection points in sequence until all inspection points are visited and finally reaches the end point E. During the visit, the ant will record the path passed and calculate the path length L ant , and use the path length as the basis for path quality evaluation;

[0036] S33. Pheromone update:

[0037] Local update: When the ant selects a path, it will perform local pheromone update on the passed path to enhance the attractiveness of this path. The local update formula is:

[0038] τ ij ←(1 - ρ)·τ ij + ρ·Δτ ij ;

[0039] Global update: After each round of iteration, the algorithm will perform global pheromone update according to the optimal path. The path with a shorter path length will release more pheromone, increasing the probability of being selected by subsequent ants. The global update formula:

[0040] τ ij←(1-ρ)·τ ij +ρ·Δτ ij ;

[0041] S34, parallel search and optimal path selection, under the multi-ant colony parallel search mechanism, multiple ant colonies execute the search process simultaneously:

[0042] S3401, multi-ant colony search: Each ant colony performs path search independently, and each ant starts from the starting point and selects the path according to the probability P ij A path is chosen until the end is reached. Pheromone updates are shared between different ant colonies, but the search paths are independent.

[0043] S3402, optimal path selection: by comparing the path qualities of all ant colonies, the route with the shortest path length is selected as the optimal path outputted in the end, which includes the shortest path passing through all patrol points and satisfies the safety constraint requirements;

[0044] S35, when the maximum number of iterations T is reached or the path optimization converges or the path quality is not further optimized, the execution is stopped;

[0045] S36, output the optimal path point sequence R = {S, p1, p2, ..., p m , E}, that is, the optimized patrol path, including the starting point, all patrol points and the end point.

[0046] Preferably, the generation process of the mark to be reviewed is as follows:

[0047] S41. Select a random number a∈(0,n) and calculate the audit code CA=g a mod q;

[0048] Among them, p = 521, which is a prime number; n = 2p-1 is a Mersenne prime number; take the corresponding p-order irreducible primitive polynomial q = x 521 +x 32 +1, x is the indefinite variable of the polynomial;

[0049] S42, select a random number k∈(0,n), calculate the first level to be examined factor F1=g k mod q;

[0050] S43, calculate the secondary factor to be examined F2 = H (List1) × k + a × F1 mod n;

[0051] Among them, List1 represents the binary string of each path point in the optimized path point set P′ cascaded in sequence;

[0052] S44, generate a pending review mark LA = {F1, F2}.

[0053] Preferably, the review process for the overall integrity of the received path point set is as follows:

[0054] S51. Calculate the first-level review factor FA1:

[0055] S52. Calculate the second-level review factor FA2:

[0056] S53. If FA1 = FA2, it indicates that the path point set P' has been completely received.

[0057] Preferably, the generation process of the patrol path points is as follows:

[0058] S61. Based on the received and optimized path point set P' = {P1, P2,..., P i ,..., P n}, connect the safety points end to end. Its core goal is:

[0059] Smooth transition between the starting point, ending point and intermediate patrol points of the path, avoiding sharp turns or large-angle offsets, and meeting the flight path smoothness of the UAV;

[0060] Consider flight constraints: the maximum turning angle of the UAV, the safety radius, and the smoothness of the flight altitude;

[0061] S62. Define problem constraints:

[0062] Safe turning angle: The turning angle between two adjacent path points needs to satisfy: θ i ≤θ max ;

[0063] Among them, θ i represents the included angle between path points p i-1 →p i and p i →p i+1 ; θ max represents the maximum allowable safe turning angle of the UAV;

[0064] Path smoothness: During the route optimization process, on the premise of ensuring less increase in path length, introduce smooth curves or transition points for connection;

[0065] Safety radius constraint: Ensure that the minimum interval between the UAV flight path points p i is greater than the safety radius R safe , to avoid unstable flight routes caused by overly dense path points;

[0066] S63. Generate the path sequence P safe after safe end-to-end connection based on the safety point end-to-end connection method: P safe={p1, p′2, ..., p′ n}。

[0067] Preferably, the implementation steps of the safe point head-to-tail connection method are as follows:

[0068] S71. For the optimized path point set P′ = {P1, P2, …, P i , …, P n}, use a third-order Bezier curve to smoothly fit the path and insert safe connection points to make the start point, end point, and intermediate point paths smoothly transition. The third-order Bezier curve formula is as follows:

[0069] B(t) = (1 - t) 3 p0 + 3t(1 - t) 2 p1 + 3t 2 (1 - t)p2 + t 3 p3;

[0070] In the formula, B(t) represents any point on the Bezier curve; p0, p1, p2, and p3 represent control points; t represents the parameter of the curve, t ∈ [0, 1];

[0071] Among them, the control point selection principle is as follows:

[0072] p0 is the start point of the path segment; p3 is the end point of the path segment; p1 and p2 are auxiliary control points, and the selection rules are as follows:

[0073] p1 is offset according to the flight direction of the start point, and the distance is λ × d i,i+1 ;

[0074] p2 is offset according to the reverse direction of the end point, and the distance is λ × d i,i+1 ;

[0075] Among them, λ ∈ (0, 1) is the smoothing coefficient;

[0076] S72. After generating the smooth path, check whether the interval between path points meets the safety radius constraint R safe , specifically as follows:

[0077] Path point distance calculation. The distance between two path points is:

[0078]

[0079] Verify the safety radius: If d i,j < R safe , then insert an additional intermediate point p k , re-divide the path, and calculate the intermediate point coordinates as:

[0080]

[0081] S73. For three consecutive path points p i-1 , p i , p i+1 , calculate the steering angle θ i :

[0082]

[0083] In the formula, represents the vector of the path segment p i-1 →p i ;

[0084] represents the vector of the path segment p i →p i+1 ;

[0085] If θ i > θ max , then adjust the path, insert transition points, and make the path steering angle meet the safety requirements.

[0086] Preferably, the generation process of the global temporary optimal route is as follows:

[0087] S81. Divide the inspection path points P safe into several clusters through the density-based clustering algorithm, and perform route merging. Divide the path points into C1, C2,..., C k clusters;

[0088] The formula of the density-based clustering algorithm is: For any path point p i , if the following conditions are met, then p i belongs to a certain cluster:

[0089] N(p i ) = {p j |d i,j ≤ ε};

[0090] In the formula, N(p i ) represents the neighborhood of the path point p i ; d i,j represents the Euclidean distance between the path points p i and p j ; ε represents the neighborhood radius, that is, the distance threshold between path points;

[0091] S82. For each clustering cluster C i , optimize the path execution order through the improved traveling salesman problem algorithm to form the shortest path. The specific operation process is as follows:

[0092] S8201. The goal is to find an optimal path sequence such that the total path length from the starting point, passing through all path points and finally returning to the starting point is minimized:

[0093]

[0094] In the formula, D TSP represents the total length of the TSP path; d i,i+1 represents the distance from path point p i to p i+1 ; n represents the number of path points within the cluster;

[0095] S8202. The path selection probability, that is, the probability that an ant moves from path point i to path point j is:

[0096]

[0097] In the formula, τ ij represents the pheromone concentration of path i→j; η ij represents the path heuristic factor; d ij represents the path length; ψ ij represents the safety constraint factor. If path i→j does not satisfy the safety radius constraint, then ψ ij <1, otherwise it is 1; α and β represent the parameters controlling the weights of pheromone and heuristic factor; N i represents the set of remaining points that the ant can currently access;

[0098] Pheromone update, that is, the pheromone concentration is updated over time, including pheromone evaporation and the contribution of the ant path:

[0099] τ ij ←(1 - ρ)·τ ij + ρ·Δτ ij ;

[0100]

[0101] In the formula, ρ represents the pheromone evaporation coefficient; Q represents the total amount of pheromone released; L k represents the total length of the path taken by ant k; m represents the number of ants;

[0102] S83. Through route clustering and dynamic path optimization, the global temporary optimal route P best = {q1, q2,…, q m} is obtained.

[0103] The technical solution of the present invention: A substation UAV dynamic route planning system, which is used to execute the above-mentioned substation UAV dynamic route planning method, includes:

[0104] A task input module, which is used to receive the input information of the inspection task and generate a set of target points according to requirements, as the basic data input for route planning;

[0105] A route planning module, which is used to divide the target points into several sub-routes based on the input inspection point data, utilize the route simplification and dynamic combination algorithm, perform minimum path optimization on each sub-route to generate a locally optimal path, and then introduce a safety point head-to-tail connection mechanism for all sub-routes, and use the dynamic programming algorithm to merge the paths of multiple sub-routes to form a complete optimal temporary route;

[0106] A safety management module, which is used to provide the safety point connection function during the flight of the UAV, automatically introduce safety points at the starting and ending points of each route according to the task requirements and the safety radius threshold, and calculate the shortest connection path between safety points during route optimization;

[0107] A UAV execution module, which is used to transmit the dynamic temporary route data generated by the route planning module to the UAV control system, control the UAV to execute the inspection task according to the planned optimal path in real time, complete the accurate arrival at the target point and data collection, and dynamically adjust the flight path according to task changes;

[0108] An AI recognition module, which is responsible for processing the image and video data collected by the UAV execution module, uploading the data to the AI analysis engine, using the target recognition algorithm to perform intelligent recognition and status analysis on the inspection target, automatically detecting equipment abnormalities, defects or safety hazards, and generating an identification result report;

[0109] A user terminal module, which is used to provide a visual operation interface for the entire task process, support the user to view the dynamic route, equipment abnormality report and the real-time status of the UAV, and at the same time quickly adjust the next inspection task plan according to the recognition result feedback to realize the man-machine interaction function of the system.

[0110] Compared with the prior art, the above technical solution of the present invention has the following beneficial technical effects:

[0111] The present invention designs a method and system for dynamic route planning of a substation UAV. By combining accurate processing of task data, route optimization and dynamic safety constraints, it ensures the efficiency, stability and safety of route planning, and effectively solves the problem of route planning in substation UAV inspections:

[0112] (1) Precise and efficient task input: By converting the inspection requirements into a detailed set of point data, combining steps such as coordinate conversion, safety radius calculation and shooting parameter setting, it ensures the generation of high-quality data that meets the inspection requirements, realizes the unified processing of multiple data sources, ensures the accuracy and reliability of coordinate data, and provides a solid foundation for subsequent route planning;

[0113] (2) Precise optimization of route planning: By using a route optimization method based on the improved ant colony algorithm (IACO), the problems of path simplification and safety constraints in UAV route planning are solved. The dynamic safety radius constraint, heuristic information guidance, and multi-ant colony parallel search mechanism are introduced, effectively improving the global optimal ability of route planning, avoiding local optimal solutions. At the same time, the adaptive pheromone update mechanism strengthens high-quality paths, and the optimization results meet the dual requirements of the shortest path and safe flight;

[0114] (3) High path safety and stability: During the route planning process, the dynamic introduction of the safety radius constraint effectively prevents the UAV from approaching obstacles or entering dangerous areas. In addition, through path smoothing, it ensures a smooth transition of the UAV between the starting point, the ending point, and the intermediate inspection points, improving the executability and flight stability of the route. And according to the goal of each route, the minimum combination calculation of the current route is carried out, that is, route simplification, with safety points formed at both the beginning and the end. Then, after each route is simplified, the safety points are connected and merged, and finally a new temporary route is formed to ultimately achieve that all observation targets are reached, also meeting the dual requirements of the shortest path and safe flight;

[0115] (4) Multi-level review mechanism to ensure data integrity: A method for generating the identification to be reviewed based on random numbers, hash functions, and path point sequences is designed, and the integrity of path point data is verified through primary and secondary review factors, effectively ensuring the accuracy and security of the optimized path data during transmission and reception, and guaranteeing the reliability of route planning;

[0116] (5) Path smoothing to improve flight quality: The path points are smoothly fitted by a third-order Bezier curve, effectively solving the problem of route instability caused by dense path points or sharp turns in path planning, ensuring the flight smoothness of the UAV under the safety radius constraint. In addition, through steering angle constraint and path point interval control, the safety and smoothness of the route are improved;

[0117] (6) High dynamic adaptability and execution efficiency: The present invention combines dynamic constraint optimization and multi-ant colony parallel search mechanism in route planning, improving the globality and efficiency of path search. At the same time, the adaptive pheromone update mechanism enables the algorithm to dynamically adjust according to different task scenarios and route requirements, ensuring the optimality and adaptability of the planning results. BRIEF DESCRIPTION OF THE DRAWINGS

[0118] Figure 1 It is a system architecture diagram of a dynamic UAV route planning system for a substation proposed by the present invention;

[0119] Figure 2 It is a method flow chart of a dynamic UAV route planning method for a substation proposed by the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0120] Example 1, as Figure 1 shown, a dynamic route planning system for a substation UAV proposed by the present invention includes a task input module, a route planning module, a safety management module, a UAV execution module, an AI recognition module, and a user terminal module.

[0121] The task input module receives the input information of the inspection task, including but not limited to the longitude and latitude of the target inspection point, the aerial photography angle, the task priority, and the safety requirements, and generates a set of target points as the basic data input for route planning according to the requirements;

[0122] Based on the input inspection point data, the route planning module uses the route simplification and dynamic combination algorithm to divide the target points into several sub-routes, performs the minimum path optimization on each sub-route to generate a locally optimal path, and then introduces the safety point head and tail connection mechanism for all sub-routes. On the premise of ensuring flight safety, the dynamic programming algorithm is used to merge the paths of multiple sub-routes to form a complete optimal temporary route;

[0123] The safety management module is used to provide the safety point connection function during the UAV flight. According to the task requirements and the safety radius threshold, safety points are automatically introduced at the starting and ending points of each route, and the shortest connection path between safety points is calculated during route optimization to ensure the UAV navigation safety and task continuity;

[0124] The UAV execution module transmits the dynamic temporary route data generated by the route planning module to the UAV control system, and controls the UAV to execute the inspection task according to the planned optimal path in real time, accurately reach the target point and collect data, and dynamically adjust the flight path according to the task changes;

[0125] The AI recognition module is responsible for processing the image and video data collected by the UAV execution module, uploading the data to the AI analysis engine, using the target recognition algorithm to perform intelligent recognition and status analysis on the inspection target, automatically detecting equipment abnormalities, defects or safety hazards, and generating an identification result report;

[0126] The user terminal module provides a visual operation interface for the entire task process, including but not limited to task input, route planning result display, UAV execution path monitoring, and AI recognition result feedback functions, supports the user to view the dynamic route, equipment abnormality report, and UAV real-time status, and can quickly adjust the next inspection task plan according to the recognition result feedback to realize the man-machine interaction function of the system.

[0127] Example 2, as Figure 2As shown in the figure, a dynamic route planning method for a substation UAV proposed by the present invention is applied to a dynamic route planning system for a substation UAV proposed in Embodiment 1, and its specific implementation steps are as follows:

[0128] S1. The task input module converts the inspection requirements into a detailed set of point data, and through steps such as coordinate conversion, safety radius calculation, and shooting parameter setting, generates high-quality data that meets the inspection requirements, laying a foundation for subsequent route planning. The specific implementation process is as follows:

[0129] S11. Obtain the geographical coordinates of the point data. The sources of the geographical coordinate data include but are not limited to:

[0130] GIS system (Geographic Information System): Provides accurate coordinate data of the equipment in the substation;

[0131] UAV flight historical data: Records known target points through historical aerial photography tasks;

[0132] Real-time detection feedback: The abnormal detection module outputs the coordinates of the fault point as a new target point;

[0133] Define the coordinates of each inspection point p i as: p i ={x i , y i , z i};

[0134] Among them, x i represents the longitude coordinate of the i-th inspection point; y i represents the latitude coordinate of the i-th inspection point; z i represents the safe inspection height of the i-th inspection point;

[0135] S12. To ensure the unity of the coordinate data, convert the data from different sources to a unified world coordinate system, and use the coordinate conversion formula: (x' i , y' i ) = T(x i , y i );

[0136] Among them, T represents the coordinate conversion function;

[0137] Accordingly, set the inspection task, and the set of inspection target points is:

[0138] P = {p1, p2,..., p i ,..., p n};

[0139] Among them, n represents the number of target points, and p i represents the i-th inspection point;

[0140] S13. To ensure the flight safety of the UAV, a safety radius δ is set for each inspection point p i : i

[0141] δ i = f(h i , r);

[0142] where h i represents the safety inspection height z of the i-th inspection point i ; r represents the positioning accuracy threshold of the UAV equipment; f represents a function for dynamically adjusting the safety radius according to the flight height and positioning accuracy, and δ i = α×h i + β×r; α represents the safety coefficient weight, that is, the influence degree of the inspection height on the safety radius; β represents the positioning accuracy weight, indicating the influence degree of the equipment accuracy on the safety radius;

[0143] S14. Set the shooting parameters for the inspection tasks of each point p i , specifically including:

[0144] Shooting angle θ i : Represents the pitch angle during UAV aerial photography;

[0145] Shooting distance d i : Represents the distance between the camera and the target equipment;

[0146]

[0147] Shooting resolution R i : Represents the resolution requirement of the aerial photography image;

[0148] S15. Accordingly, output the set of safety inspection points: P = {(x′ i , y′ i , z i , δ i , θ i , d i , R i )|i = 1, 2,..., n}.

[0149] S2. The route planning module, to optimize the route planning of the UAV during substation inspections, through the route simplification algorithm, proposes a method based on the Improved Ant Colony Optimization (IACO). Combining heuristic information and dynamic constraint conditions, through the path selection probability, dynamic safety constraints, and multi-ant colony parallel search mechanism, it completes the optimization and simplification of the inspection point paths and finally generates the optimal route. The specific implementation process is as follows: ​

[0150] S21. Input data: Set of safety inspection points: P = {(x′ i , y′ i , z i , δ i , θ i , d i , R i ) | i = 1, 2,..., n};

[0151] Define the starting safety point S = (x′0, y′0, z0);

[0152] Define the ending safety point E = (x′ n , y′ n , z n );

[0153] S22. Objective: Plan the shortest flight path R passing through all inspection points, minimize the path length to the greatest extent while ensuring path safety, and output the sequence of optimized flight path points and the total path length;

[0154] S23. Dynamic path constraint optimization: Introduce a safety radius constraint during path selection to prevent the UAV from approaching obstacles or entering dangerous areas. The path selection probability P ij The formula is:

[0155]

[0156] In the formula, τ ij represents the pheromone concentration of path i → j; η ij represents the path heuristic factor; d ij represents the path length; ψ ij represents the safety constraint factor. If path i → j does not meet the safety radius constraint, then ψ ij < 1, otherwise it is 1; α and β represent the parameters controlling the weights of pheromone and heuristic factor; U represents the set of remaining points that the ant can currently access;

[0157] S24. Multi - ant - colony parallel search mechanism: Introduce multiple ant colonies to execute search tasks in parallel, increase the global search ability of the algorithm, and effectively avoid the problem of local optimal solutions. Each ant colony searches for paths independently and finally selects the global optimal path for output;

[0158] S25. Adaptive pheromone update mechanism: Dynamically adjust the pheromone update amount according to the path quality, strengthen the better paths and weaken the inferior paths:

[0159] τ ij ← (1 - ρ)·τ ij + ρ·Δτ ij ;

[0160]

[0161] Wherein, ρ represents the pheromone evaporation coefficient; Q represents the pheromone intensity; L represents the path length; f adj represents the dynamic adjustment factor, which is adaptively adjusted based on the current iteration round and path quality;

[0162] S26. The execution process of the method based on the Improved Ant Colony Optimization (IACO) is as follows:

[0163] S2601. Initialization: Input the inspection point data P, initialize the number of ants m, the maximum number of iterations T, the pheromone concentration τ ij , the heuristic factor η ij and set the safety radius constraint δ i ;

[0164] S2602. Ant path search:

[0165] Each ant starts from the starting point S and visits the inspection points in sequence based on the path selection probability P ij according to the formula:

[0166] (1) Calculate the selection probability P ij of each optional path, and select the next inspection point to visit based on this probability;

[0167] (2) Selection strategy: The ant selects the next point according to the probability P ij of the current path. If there are multiple selectable paths, it is more inclined to select the path with a higher pheromone concentration;

[0168] (3) Safety constraint: The selected path d ij needs to meet the minimum safety radius δ i requirement to ensure the flight safety of the drone;

[0169] Path execution and recording: Each ant passes through multiple inspection points in sequence until all inspection points are visited and finally reaches the end point E. During the visit, the ant will record the path it passes through and calculate the path length L ant , and use the path length as the basis for path quality evaluation;

[0170] S2603. Pheromone update:

[0171] (1) Local update: When the ant selects a path, it will perform local pheromone update on the path it passes through to enhance the attractiveness of this path. The local update formula is:

[0172] τ ij ←(1 - ρ)·τij +ρ·Δτ ij ;

[0173] (2) Global update: After each iteration ends, the algorithm will perform global pheromone update according to the optimal path. Paths with shorter lengths will release more pheromones, increasing the probability of being selected by subsequent ants. The global update formula:

[0174] τ ij ←(1 - ρ)·τ ij +ρ·Δτ ij ;

[0175] S2604. Parallel search and optimal path selection:

[0176] Under the multi - ant - colony parallel search mechanism, multiple ant colonies execute the search process simultaneously:

[0177] (1) Multi - ant - colony search: Each ant colony independently executes path search. Each ant starts from the starting point and selects a path according to the path selection probability P ij until it reaches the end point. Pheromone updates are shared among different ant colonies, but the search paths are independent;

[0178] (2) Optimal path selection: By comparing the path qualities of all ant colonies, select the flight route with the shortest path length as the optimal path for the final output. This optimal path includes the shortest path passing through all inspection points and meets the safety constraint requirements;

[0179] S2605. Stop execution when the maximum number of iterations T is reached, or the path optimization converges, or the path quality cannot be further optimized;

[0180] S2606. Output the optimal path point sequence R = {S, p1, p2, …, p m , E}, that is, the inspected path after optimization, including the starting point, all inspection points, and the end point;

[0181] The total path D, that is, the total length of the optimal path,

[0182] S27. Output the optimized path point set P′ = {P1, P2, …, P i , …, P n};

[0183] Among them, P1 is the starting point, P k is the end point, and P i (i = 2, …, k - 1) are the intermediate refined inspection points;

[0184] S28. Generate a to - be - reviewed identifier for the optimized path point set P′. The specific implementation process is as follows:

[0185] S2801. Select a random number \(a\in(0,n)\) and calculate the verification code \(C_A = g^{a}\bmod q\); a \(\bmod q\);

[0186] where \(p = 521\), a prime number; \(n=2^{p}-1\) is a Mersenne prime number; take the corresponding \(p\)-th irreducible primitive polynomial \(q=x^{p}+x^{6}+1\), where \(x\) is a polynomial indeterminate; p Take the corresponding \(p\)-th irreducible primitive polynomial \(q = x^{p}+x^{6}+1\), \(x\) is a polynomial indeterminate; 521 \(+x^{6}\) 32 \(+1\), \(x\) is a polynomial indeterminate;

[0187] S2802. Select a random number \(k\in(0,n)\) and calculate the first-level factor to be reviewed \(F_1 = g^{k}\bmod q\); k \(\bmod q\);

[0188] S2803. Calculate the second-level factor to be reviewed \(F_2=H(List1)\times k + a\times F_1\bmod n\);

[0189] where \(List1\) represents the binary string obtained by concatenating each path point in the optimized path point set \(P'\) in order; \(H\) represents a hash function;

[0190] S2804. Generate the identification to be reviewed \(L_A=\{F_1,F_2\}\);

[0191] S29. Transmit \(\{L_A,P'\}\) to the security management module.

[0192] S3. The security management module outputs a safe and efficient flight path by accurately calculating the path distance, checking the safety constraints, adjusting the path to bypass the safe area, and ensuring the smoothness of the flight through smoothing processing. The specific implementation process is as follows:

[0193] S31. Based on the identification to be reviewed \(L_A\), review the overall integrity of the received path point set. The review process is as follows:

[0194] S3101. Calculate the first-level review factor \(F_{A1}\):

[0195] S3102. Calculate the second-level review factor \(F_{A2}\):

[0196] S3103. If \(F_{A1}=F_{A2}\), it means that the path point set \(P'\) has been completely received;

[0197] S32. Based on the received optimized path point set \(P'=\{P_1,P_2,\cdots,P_m,\cdots,P_n\}\), connect the safety points end to end. The core goal is: i \(\cdots,P_m\) n \(\cdots,P_n\}\), connect the safety points end to end. The core goal is:

[0198] Smoothly transition between the starting point, ending point, and intermediate inspection points of the path to avoid sharp turns or large-angle offsets, and meet the smoothness of the UAV's flight path;

[0199] Consider flight constraints: the maximum turning angle of the UAV, the safety radius, and the smoothness of the flight altitude;

[0200] S33. Define problem constraints:

[0201] S3301. Safe turning angle: The turning angle between two adjacent path points should satisfy: θ i ≤θ max ;

[0202] Among them, θ i represents the included angle between path points p i-1 →p i and p i →p i+1 ; θ max represents the maximum allowable safe turning angle of the UAV;

[0203] S3302. Path smoothness: During the route optimization process, introduce a smooth curve or transition points for connection on the premise of ensuring less increase in path length;

[0204] S3303. Safety radius constraint: Ensure that the minimum interval between the UAV flight path points p i is greater than the safety radius R safe , to avoid unstable flight routes caused by overly dense path points;

[0205] S34. Accordingly, design a method for connecting the beginning and end of safety points, and its specific implementation process is as follows:

[0206] S3401. For the optimized path point set P′ = {P1, P2,..., P i ,..., P n}, use a third-order Bezier curve to smoothly fit the path and insert safety connection points to make the starting point, ending point, and intermediate point paths smoothly transition. The third-order Bezier curve formula is as follows:

[0207] B(t) = (1 - t) 3 p0 + 3t(1 - t) 2 p1 + 3t 2 (1 - t)p2 + t 3 p3;

[0208] In the formula, B(t) represents any point on the Bezier curve; p0, p1, p2, and p3 represent control points; t represents the parameter of the curve, t ∈ [0, 1];

[0209] Among them, the control point selection principle is as follows:

[0210] p0 is the starting point of the path segment; p3 is the ending point of the path segment; p1 and p2 are auxiliary control points, and the selection rules are as follows:

[0211] p1 is offset according to the flight direction of the starting point, and the distance is λ×d i,i+1 ;

[0212] p2 is offset according to the reverse direction of the ending point, and the distance is λ×d i,i+1 ;

[0213] where λ∈(0,1) is the smoothing coefficient;

[0214] S3402. After generating the smooth path, check whether the interval between path points meets the safety radius constraint R safe , specifically as follows:

[0215] Calculation of the distance between path points. The distance between two path points is:

[0216]

[0217] Verify the safety radius: If d i,j <R safe , then insert an additional intermediate point p k , re-divide the path, and the coordinates of the intermediate point are calculated as:

[0218]

[0219] It should be noted that inserting the intermediate point p k smoothly divides the overly dense path points, gradually increasing the distance between path points, and generating new transition points p k through linear interpolation, enabling the UAV to smoothly transition from p i to p j during flight, thereby reducing the severity of flight turning. Inserting the intermediate point p k can re-divide the path to ensure that the interval d i,k between every two adjacent points and d k,j both meet d i,j ≧R safe , thereby meeting the safety radius constraint;

[0220] S3403. For three consecutive path points p i-1 、p i 、p i+1 , calculate the turning angle θ i :

[0221]

[0222] Wherein, represents the path segment p i-1 →p i vector;

[0223] represents the path segment p i →p i+1 vector;

[0224] If θ i >θ max , then adjust the path, insert transition points, and make the turning angle of the path meet the safety requirements;

[0225] S35. Generate the path sequence P after safe head-to-tail connection safe : P safe ={p1, p′2,..., p′ n};

[0226] The total path length D safe :

[0227] S36. Transmit the path sequence P after safe head-to-tail connection safe to the route planning module.

[0228] S4. The route planning module merges the route points through the DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering algorithm and optimizes the path order using the improved ant colony algorithm, achieving the merging and optimization of the dynamic route. The specific implementation process is as follows:

[0229] S41. Divide the inspection path points P safe into several clusters through the DBSCAN algorithm and perform route merging, dividing the path points into C1, C2,..., C k clusters;

[0230] The DBSCAN algorithm formula is: For any path point p i , if the following conditions are met, then p i belongs to a certain cluster:

[0231] N(p i )={p j |d i,j ≤ε};

[0232] Wherein, N(p i ) represents the neighborhood of the path point p i ; d i,j represents the distance between the path point p i and p jThe Euclidean distance between; ε represents the neighborhood radius, i.e., the distance threshold between path points;

[0233] S42. For each cluster C i , the execution order of the path is optimized by an improved Traveling Salesman Problem (TSP) algorithm to form the shortest path, and the specific operation process is as follows:

[0234] S4201. The goal is to find an optimal path sequence such that the total path length from the starting point, passing through all path points, and finally returning to the starting point is minimized:

[0235]

[0236] In the formula, D TSP represents the total length of the TSP path; d i,i+1 represents the distance from path point p i to p i+1 ; n represents the number of path points within the cluster;

[0237] S4202. The path selection probability, that is, the probability that an ant moves from path point i to path point j is:

[0238]

[0239] In the formula, τ ij represents the pheromone concentration of path i→j; η ij represents the path heuristic factor; d ij represents the path length; ψ ij represents the safety constraint factor. If path i→j does not satisfy the safety radius constraint, then ψ ij <1, otherwise it is 1; α and β represent the parameters that control the weights of pheromone and heuristic factor; N i represents the set of remaining points that the ant can currently access;

[0240] Pheromone update, that is, the pheromone concentration is updated over time, including pheromone evaporation and the contribution of the ant path:

[0241] τ ij ←(1 - ρ)·τ ij + ρ·Δτ ij ;

[0242]

[0243] In the formula, ρ represents the pheromone evaporation coefficient; Q represents the total amount of pheromone released; L k represents the total length of the path taken by ant k; m represents the number of ants;

[0244] S43. Obtain the globally temporarily optimal route P through route clustering and dynamic path optimization best ={q1, q2,…, q m};

[0245] Total path length D best :

[0246] S44. Transmit the globally temporarily optimal route P best to the UAV execution module.

[0247] S5. The UAV execution module uploads the globally temporarily optimal route P best to the UAV control system. The UAV flies in sequence according to the route, visits all inspection target points, and performs image acquisition and task execution at each target point p i .

[0248] S6. The UAVs performing the inspection task transmit the collected images and data to the AI recognition module. The AI recognition module performs the following tasks: image processing, anomaly detection, and generating an analysis report, outputs the recognition results, and based on this, the user terminal module visualizes the planned route, the collected images, and the recognition results.

[0249] The embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings. However, the present invention is not limited to this. Various changes can be made without departing from the spirit of the present invention within the knowledge scope of those skilled in the art to which the present invention pertains.

Claims

1. A dynamic route planning method for UAVs in a substation, characterized in that, It includes the following specific implementation steps: S1. Convert the inspection requirements into a detailed set of point location data, and through steps such as coordinate transformation, safety radius calculation, and shooting parameter setting, generate an output set of safe inspection point locations that meet the inspection requirements; S2. Combine heuristic information and dynamic constraint conditions, and through path selection probability, dynamic safety constraints, and multi-ant colony parallel search mechanism, complete the optimization and streamlining of the inspection point path, output the optimized path point set, and generate a to-be-reviewed identifier for the optimized path point set; S3. Based on the to-be-reviewed identifier, review the overall integrity of the received path point set. Based on the optimized path point set, use the third-order Bezier curve for path smoothing fitting to ensure smooth transitions between the starting point, ending point, and intermediate points. Consider flight constraints during path smoothing, including maximum turning angle, path smoothness, and safety radius, insert safety connection points and intermediate points to ensure that the interval between every two path points meets the safety radius requirements, and adjust the turning angle to meet the maximum safe turning angle constraint, and finally generate the inspection path points; S4. Use the density-based clustering algorithm to perform spatial clustering on the path points, divide the path points into multiple clusters, and for each cluster, use the improved TSP algorithm to optimize the path order to minimize the total path length. The path selection is dynamically adjusted according to pheromone concentration, heuristic factor, path length, and safety constraint factor. Through clustering and path optimization, obtain the globally temporarily optimal flight path; S5. Upload the globally temporarily optimal flight path to the UAV control system, and the UAV flies in sequence according to the flight path, visits all inspection target point locations, and performs image acquisition and task execution at each target point location; S6. Use the collected images and data for image processing, anomaly detection, and generate an analysis report, output the recognition results, and visualize the planned flight path, the collected images, and the recognition results accordingly.

2. The method for dynamically planning the flight path of an unmanned aerial vehicle for a substation according to claim 1, wherein The generation process of the optimized path point set is as follows: S21. Input data: Safety inspection point set: P = {(x' i , y' i , z i , δ i , θ i , d i , R i ) | i = 1, 2,..., n}; Define the starting safety point S=(x'0, y'0, z0); Define the termination safe point E = (x' n , y' n , z n ); where n represents the number of safe inspection point locations; S22. Objective: Plan the shortest flight path R that passes through all inspection points, and minimize the path length to the greatest extent while ensuring path safety, and output the streamlined and optimized flight path point sequence and the total path length; S23. Dynamic path constraint optimization. Introduce a safety radius constraint during the path selection process to prevent the UAV from approaching obstacles or entering dangerous areas. The path selection probability P ij The formula is as follows: where τ ij represents the pheromone concentration of the path i→j; η ij represents the path heuristic factor; d ij represents the path length; ψ ij represents the safety constraint factor. If the path i→j does not satisfy the safety radius constraint, then ψ ij < 1, otherwise it is 1; α and β represent the parameters that control the weights of pheromone and heuristic factor; U represents the set of remaining points that the ant can currently access; S24. Multi-ant colony parallel search mechanism: Introduce multiple ant colonies to execute search tasks in parallel, increase the global search ability of the algorithm, effectively avoid the problem of local optimal solutions, each ant colony independently searches for paths, and finally selects the globally optimal path for output; S25. Adaptive pheromone update mechanism, dynamically adjust the pheromone update amount according to the path quality, strengthen the better paths, and weaken the inferior paths: τ ij ←(1 - ρ)·τ ij +ρ·Δτ ij ; where ρ represents the pheromone evaporation coefficient; Q represents the pheromone intensity; L represents the path length; f adj represents the dynamic adjustment factor, which is adaptively adjusted based on the current iteration round and path quality; S26. Output the optimized path point set based on the improved ant colony algorithm.

3. The method for dynamically planning the flight path of an unmanned aerial vehicle for a substation according to claim 2, wherein, The process of outputting the optimized path point set based on the improved ant colony algorithm is as follows: S31. Initialization: Input the inspection point data P, initialize the number of ants m, the maximum number of iterations T, the pheromone concentration τ ij , the heuristic factor η ij and set the safety radius constraint δ i ; S32. Ant path search: Each ant starts from the starting point S and visits the inspection points in sequence according to the path selection probability P ij according to the formula: S3201. Calculate the selection probability P of each optional path ij , and select the next inspection point to visit based on this probability; S3202. Selection strategy: The ant selects the next point according to the probability P of the current path. ij If there are multiple paths to choose from, it is more inclined to choose the path with a higher pheromone concentration. S3203. Safety Constraint: Selected path d ij shall satisfy the minimum safety radius δ i to ensure the flight safety of the UAV; Path execution and recording: Each ant sequentially passes through multiple inspection points until all inspection points have been visited and finally reaches the end point E. During the visit, the ant records the path it has passed through and calculates the path length L ant , and uses the path length as the basis for path quality evaluation; S33. Pheromone update: Local update: When an ant selects a path, it will perform local pheromone update on the path it has passed through to enhance the attractiveness of this path. The local update formula is: τ ij ←(1 - ρ)·τ ij +ρ·Δτ ij ; Global update: After each iteration, the algorithm performs global pheromone update according to the optimal path. Paths with shorter lengths will release more pheromones, increasing the probability of being selected by subsequent ants. The global update formula: τ ij ←(1 - ρ)·τ ij +ρ·Δτ ij ; S34. Parallel search and optimal path selection. Under the multi-ant colony parallel search mechanism, multiple ant colonies perform the search process simultaneously: S3401. Multi - ant - colony search: Each ant colony independently performs path search. Each ant starts from the starting point and selects a path according to the path - selection probability P ij to select a path until it reaches the end point. Pheromone updates are shared among different ant colonies, but the search paths are independent; S3402. Optimal path selection: By comparing the path qualities of all ant colonies, select the flight route with the shortest path length as the optimal path for the final output. This optimal path includes the shortest path passing through all inspection points and meets the safety constraint requirements at the same time; S35. Stop execution when the maximum number of iterations T is reached, or the path optimization converges, or the path quality is not further optimized; S36. Output the optimal path point sequence R = {S, p1, p2, …, p m , E}, that is, the optimized inspection path, including the starting point, all inspection points, and the ending point.

4. A dynamic route planning method for an unmanned aerial vehicle in a substation according to claim 1, characterized in that The generation process of the to-be-reviewed identifier is as follows: S41. Select a random number a ∈ (0, n), and calculate the verification code CA = g a mod q; Among them, p = 521, which is a prime number; n = 2 p -1 is a Mersenne prime number; take the corresponding p-th irreducible primitive polynomial q = x 521 + x 32 + 1, where x is a polynomial indeterminate; S42. Select a random number k ∈ (0, n), and calculate the first-level factor to be reviewed F1 = g k mod q; S43. Calculate the secondary to-be-reviewed factor F2 = H(List1) × k + a × F1 mod n; Among them, List1 represents the binary string obtained by concatenating each path point in the optimized path point set P' in sequence; S44. Generate the to-be-reviewed identifier LA = {F1, F2}.

5. A dynamic route planning method for a substation UAV according to claim 1, characterized in that The review process for receiving the overall integrity of the path point set is as follows: S51. Calculate the first-level review factor FA1: S52. Calculate the secondary review factor FA2: S53. If FA1 = FA2, it means that the path point set P' has been completely received.

6. The dynamic route planning method for a substation UAV according to claim 1, wherein The generation process of the inspection path points is as follows: S61. Based on the received optimized set of path points P' = {P1, P2, …, P i , …, P n}, connect the safety points end to end. Its core objective is: Smooth transition between the starting point, ending point and intermediate inspection points of the path to avoid sharp turns or large-angle offsets, meeting the flight path smoothness of the UAV; Consider flight constraints: the maximum turning angle of the UAV, the safety radius and the smoothness of the flight altitude; S62. Define problem constraints: Safe steering angle: The steering angle between two adjacent path points should satisfy: θ i ≤ θ max ; Among them, θ i represents the angle between the path points p i-1 →p i and p i →p i+1 ; θ max represents the maximum safe steering angle allowed for the drone; Path smoothness: During the flight route optimization process, on the premise of ensuring less increase in path length, introduce smooth curves or transition points for connection; Safety radius constraint: Ensure that the minimum distance between the flight path points p of the UAV i is greater than the safety radius R safe , to avoid unstable flight routes caused by overly dense path points; S63. Generate the path sequence P after safe head-to-tail connection based on the safe head-to-tail connection method safe : P safe = {p1, p'2,..., p' n} 7. A method for dynamic route planning of an unmanned aerial vehicle in a substation according to claim 6, characterized in that, The implementation steps of the safety point end-to-end connection method are as follows: S71. For the optimized path point set P' = {P1, P2, …, P i , …, P n}, a third-order Bézier curve is used to smoothly fit the path, and safety connection points are inserted to enable smooth transitions at the starting point, ending point, and intermediate points. The formula for the third-order Bézier curve is as follows: B(t) = (1 - t) 3 p0 + 3t(1 - t) 2 p1 + 3t 2 (1 - t)p2 + t 3 p3; In the formula, B(t) represents any point on the Bezier curve; p0, p1, p2 and p3 represent control points; t represents the parameter of the curve, t ∈ [0,1]; Among them, the control point selection principle is as follows: p0 is the starting point of the path segment; p3 is the ending point of the path segment; p1 and p2 are auxiliary control points, and the selection rules are as follows: p1 is offset according to the flight direction from the starting point, and the distance is λ × d i,i+1 ; p2 is offset in the reverse direction of the end point by a distance of λ × d i,i+1 ; Among them, λ ∈ (0,1) is the smoothness coefficient; S72. After generating the smooth path, check whether the interval between path points meets the safety radius constraint R safe , as follows: Path point distance calculation. The distance between two path points is: Verification safety radius: If d i,j < R safe , then insert an additional intermediate point p k , re-divide the path, and calculate the intermediate point coordinates as: S73. For three consecutive path points p i-1 , p i , p i+1 , calculate the turning angle θ i : In the formula, represents the path segment p i-1 →p i vector; Denote the path segment p i →p i+1 vector; If θ i > θ max , the path is adjusted and transition points are inserted to make the turning angle of the path meet the safety requirements.

8. The method for dynamically planning an unmanned aerial vehicle route in a substation according to claim 1, wherein The generation process of the global temporary optimal flight route is as follows: S81. Divide the inspection path points P into several clusters through a density-based clustering algorithm, and perform route merging, dividing the path points into clusters C1, C2, …, C safe ; k cluster; The formula of the density-based clustering algorithm is: for any path point p i , if the following conditions are met, then p i belongs to a certain cluster: N(p i ) = {p j |d i,j ≤ ε}; where N(p i ) represents the neighborhood of the path point p i ; d i,j represents the Euclidean distance between the path points p i and p j ; ε represents the neighborhood radius, i.e., the distance threshold between path points; S82. For each clustering cluster C i , the order of path execution is optimized by an improved traveling salesman problem algorithm to form the shortest path. The specific operation process is as follows: S8201. The goal is to find an optimal path sequence such that the total path length from the starting point, passing through all path points and finally returning to the starting point is minimized: Where D TSP represents the total length of the TSP path; d i,i+1 represents the distance between path points p i and p i+1 ; n represents the number of path points within the cluster; S8202. Path selection probability, that is, the probability that an ant moves from path point i to path point j is: where τ ij represents the pheromone concentration of the path i→j; η ij represents the path heuristic factor; d ij represents the path length; ψ ij represents the safety constraint factor. If the path i→j does not satisfy the safety radius constraint, then ψ ij <1, otherwise it is 1; α and β represent the parameters that control the weights of pheromone and heuristic factor; N i represents the set of remaining points that the ant can currently access; Pheromone update, that is, the pheromone concentration is updated over time, including pheromone evaporation and the contribution of the ant path: τ ij ← (1 - ρ)·τ ij + ρ·Δτ ij ; where ρ represents the pheromone evaporation coefficient; Q represents the total amount of pheromone released; L k represents the total length of the path taken by ant k; m represents the number of ants; S83. Obtain the globally temporarily optimal route P through route clustering and dynamic path optimization best ={q1, q2, …, q m}.

9. A dynamic route planning system for a substation UAV, which is used to execute the dynamic route planning method for a substation UAV according to any one of claims 1 to 8, characterized in that, Including: The task input module is used to receive the input information of the inspection task and generate a set of target points according to the requirements as the basic data input for flight route planning; The route planning module is used to divide the target points into several sub-routes based on the input inspection point data, utilize the route simplification and dynamic combination algorithm, perform minimum path optimization on each sub-route to generate a locally optimal path, and then introduce a safety point head-to-tail connection mechanism for all sub-routes, and use the dynamic programming algorithm to merge the paths of multiple sub-routes to form a complete optimal temporary route; The safety management module is used to provide the safety point connection function during the UAV flight process. According to the task requirements and the safety radius threshold, safety points are automatically introduced at the starting and ending points of each route, and the shortest connection path between safety points is calculated during route optimization; The UAV execution module is used to transmit the dynamic temporary route data generated by the route planning module to the UAV control system, and in real-time control the UAV to perform the inspection task according to the planned optimal path, complete the accurate arrival at the target point and data collection, and dynamically adjust the flight path according to task changes; The AI recognition module is responsible for processing the image and video data collected by the UAV execution module, uploading the data to the AI analysis engine, using the target recognition algorithm to perform intelligent recognition and status analysis on the inspection target, automatically detecting equipment abnormalities, defects or safety hazards, and generating an identification result report; The user terminal module is used to provide a visual operation interface for the entire task process, support the user to view the dynamic route, equipment abnormality report and the real-time status of the UAV, and at the same time quickly adjust the next inspection task plan according to the recognition result feedback to realize the man-machine interaction function of the system.

Images