Power inspection path generation method, device and equipment

Abstract

The invention discloses an electric power inspection path generation method, device and equipment, and the method comprises the steps: loading off-line map information covered by electric power inspection according to an operation flight task of an electric power inspection unmanned plane, and obtaining electric power inspection mark map information, the operation flight task comprising a take-off point position and each electric power inspection task point position; determining an electric power inspection flight path based on the electric power inspection mark map information, wherein the electric power inspection flight path is a path which takes off from the take-off point position and passes through each electric power inspection task point position; constructing a local inspection space between the current position of the inspection unmanned aerial vehicle and the local target point; in the local inspection space, an inspection unmanned aerial vehicle carries a sensor to detect sudden obstacle information; obtaining a real-time local path adjustment strategy based on the sudden obstacle information; and performing real-time path adjustment on the power inspection flight path based on a real-time local path adjustment strategy.

Description

Technical Field

[0001] This specification relates to the field of path planning technology, and in particular to a method, device, and equipment for generating a power inspection path. Background Art

[0002] In the field of power inspection, traditional power inspection mainly relies on manual climbing, telescope observation or helicopter cruise. These methods are not only costly and inefficient, but also increasingly limited in complex scenarios such as high-voltage transmission lines, substations and remote mountainous areas. The limitations of manual inspection are increasingly prominent, and there are problems such as high safety risks and insufficient data collection accuracy.

[0003] In recent years, with the maturation of drone and communication technologies, drone technology, with its advantages such as flexibility, wide operating range, low cost, and close-range observation capabilities, has gradually become a key tool in the field of power inspection, replacing manual inspections and gaining favor with power companies and research institutions. However, in complex and ever-changing inspection scenarios, drones face the daunting challenge of dynamic path planning, necessitating an urgent need to overcome existing technological bottlenecks to achieve more efficient and safe autonomous inspections. Summary of the Invention

[0004] One or more embodiments of this specification provide a method, device, and apparatus for generating a power inspection path, which are used to solve the technical problems raised in the background art.

[0005] One or more embodiments of this specification adopt the following technical solutions:

[0006] One or more embodiments of this specification provide a method for generating a power inspection path, the method comprising:

[0007] Loading offline map information covered by the power inspection according to the operational flight mission of the power inspection UAV to obtain power inspection marked map information, wherein the operational flight mission includes the take-off point location and the locations of each power inspection task point;

[0008] Determine a power inspection flight path based on the power inspection mark map information, where the power inspection flight path takes off from the take-off point and passes through each of the power inspection task points;

[0009] Construct a local inspection space between the current position of the inspection drone and the local target point;

[0010] Detecting sudden obstacle information in the local inspection space by using sensors carried by the inspection drone;

[0011] Based on the sudden obstacle information, a real-time local path adjustment strategy is obtained;

[0012] The power inspection flight path is adjusted in real time based on the real-time local path adjustment strategy.

[0013] It should be noted that unpredictable and sudden obstacles (such as wild animals or severe weather changes) often appear in power inspection areas (such as high-voltage transmission lines or remote mountainous areas). These obstacles will interrupt the planned flight path, increase the risk of collision or cause the inspection mission to fail. This method can instantly identify these interference factors and generate adjustment strategies through the construction of local inspection space and real-time sensor detection. Through this dynamic adjustment mechanism, the inspection path can bypass obstacles, maintain the continuous flight and stable observation of the drone, and thus directly improve the safety and execution efficiency of the mission. This avoids the need for manual intervention or mission termination, ensures the reliability of the inspection process and the integrity of data collection, and ultimately achieves more efficient autonomous inspections in complex and changing environments, reducing overall operational risks and optimizing resource utilization.

[0014] Furthermore, the determining of the power inspection flight path based on the power inspection mark map information includes:

[0015] Setting parameter information of the ant lion algorithm, including the number of ant populations, the number of ant lion populations, the number of iterations, and the dimension, wherein the dimension includes the three-dimensional coordinates of multiple path points generated by individual ants and ant lions in space;

[0016] Initializing parameter information of the ant lion algorithm through a pre-set chaotic map to generate path points of ant individuals and ant lion individuals in space;

[0017] Smoothing the path points to generate a path point set;

[0018] A power inspection flight path is generated based on the set of path points.

[0019] It should be noted that in complex three-dimensional scenarios such as high-voltage transmission line inspections, path planning must balance geographical constraints (e.g., avoiding mountains and vegetation), equipment accessibility (e.g., proximity to insulator strings), and flight efficiency. Traditional planning methods are susceptible to sudden changes in terrain or densely packed equipment layouts, resulting in risks such as sharp turns, ineffective circling, and proximity to obstacles. This method utilizes chaotic mapping to optimize the initialization process of the ant lion algorithm. Essentially, it uses nonlinear mapping to improve the spatial traversability and uniformity of the population's initial positions. Combined with path point smoothing, the distribution of the generated initial path points in three-dimensional space is more adapted to the layout characteristics of power facilities (e.g., natural extension along the line direction and smooth transitions). This not only avoids the problem of random initialization easily falling into local inferior solutions, but also forms a continuous and smooth spatial curve. This basic path characteristic enables subsequent UAVs to more naturally adapt to the spatial distribution characteristics of inspection targets during inspections, reducing the risk of control instability, equipment observation angle deviation, and excessive maneuvering time loss caused by sharp path fluctuations or frequent sharp turns, thus laying the foundation for efficient and stable autonomous inspections by UAVs in complex geographical environments.

[0020] Furthermore, the parameter information of the ant lion algorithm is initialized by a preset chaotic map to generate path points of ants and ant lions in space, including:

[0021] pass Initialize the parameter information of the ant lion algorithm and generate the path points of ant individuals and ant lion individuals in space Among them, d i With e i are all random numbers (0,1), u is the control parameter, w i+1 is the iterative chaotic sequence value, (x n,m ,y n,m ,z n,m ) is the coordinate of the nth ant lion individual, the mth path point along the X, Y, and Z axes, n is the maximum population size, and m is the number of path points included in the set ant lion individual.

[0022] Furthermore, the smoothing process is performed on the path points to generate a path point set, including:

[0023] The path points are smoothed using the cubic spline interpolation method to associate adjacent path points in the flight path of the power inspection drone. The resulting multiple interpolation points and the starting and ending points constitute a track point set, which are connected to obtain the flight path set of the power inspection drone.

[0024] It's important to note that in scenarios with densely distributed power equipment (such as substations or transmission line corridors), while the basic path points generated by the Ant Lion algorithm avoid major obstacles, the spatial jumps between discrete points mean that connecting them directly with straight lines will result in a zigzag path with sharp turns. This path forces the drone to frequently accelerate and decelerate, making sharp turns, which not only increases motor energy consumption and control burdens, but also causes the camera to lose precise focus on key inspection targets (such as cracks in insulator strings or conductor connection points) due to severe shaking. Cubic spline interpolation, on the other hand, uses mathematical constraints to automatically generate a continuous, smooth curvature transition curve between adjacent path points. This curve possesses the core property of second-order derivative continuity, meaning that the drone's steering angular velocity and acceleration remain constant as it flies along this curve. This results in a dual optimization: on the flight side, by eliminating the risk of sudden changes in attitude, the flight control system can maintain a stable flight attitude, avoiding sudden turns or positioning drift. On the observation side, the onboard detection equipment can maintain its optical axis aligned with the target while flying at a constant speed and low jitter, ensuring effective image acquisition time and image clarity at each inspection point. This flight smoothness, derived from mathematical continuity, essentially lays an "invisible track" for drones in complex space operations, enabling them to balance safety and precision when performing detailed inspection tasks, ultimately achieving high-quality autonomous inspections with full coverage of equipment.

[0025] Furthermore, generating a power inspection flight path based on the set of path points includes:

[0026] Input the set of path points into a pre-generated path evaluation fitness function to obtain the fitness value of each path point, and mark the minimum fitness value as the candidate ant lion;

[0027] In the current iteration, the Levy flight strategy is applied to each ant in the ant population, and the coordinates of the ant's path points in space are updated based on the influence of the position of the ant lion randomly selected in a preset manner and the position of the candidate ant lion.

[0028] The randomly selected ant lion individual and the candidate ant lion are processed by the elite pool mechanism and the Gauss-Cauchy hybrid mutation method to obtain a processing result;

[0029] Calculating the fitness values ​​of the individual ants and the individual ant lions after the wandering is completed based on the processing results;

[0030] The ant population and the ant lion population are sorted based on the fitness value, and the individual with the smallest fitness value in the historical iteration is marked as an elite ant lion.

[0031] It should be noted that in power line scenarios with interlaced mountainous areas and canyons, the drone path needs to avoid obstacles in three dimensions (such as avoiding protruding ridges, high-voltage towers, and meteorological equipment), while also meeting the viewing angle adjustment requirements for fine inspections (such as controlling the close observation angle of insulator strings). Traditional optimization algorithms are prone to falling into suboptimal paths due to dimensionality explosion or local terrain traps (such as repeatedly circling the same valley or excessively raising the route, resulting in observation failure). This method first uses the long and short jump characteristics of Levy flight (random step length based on power-law distribution) to enable individual ants to have both large-scale exploration and small-step fine-tuning capabilities when following the movement of the ant lion. This means that when a cluster of pathpoints falls into a local depression (such as surrounded by a ring of mountains), long jump steps can directly cross the terrain barrier to explore new areas, avoiding the group's repeated consumption of computing power in invalid areas. Simultaneously, the elite pool mechanism continuously retains the optimal fitness solution (candidate antlions) of each generation and applies perturbations of varying intensities to elite individuals through Gaussian-Cauchy hybrid mutation. Gaussian mutation fine-tunes coordinate points to optimize subtle flight postures (such as translating pathpoints to aim the camera at insulator hardware), while Cauchy mutation confers the ability to make large coordinate jumps (such as jumping directly from one side of a valley to the other to avoid an entire thunderstorm area). This strategy of "gradual optimization and radical breakthrough" coexists, inheriting the genes of historical high-quality paths while actively injecting diversity. Finally, using the path evaluation function (which integrates weights such as flight distance, safety margin, and observation angle) as the screening criterion, the elite antlion individuals formed after multiple rounds of iteration naturally represent a path set that naturally combines the comprehensive advantages of obstacle avoidance safety, movement efficiency, and observation integrity. Essentially, this algorithm structure "pre-positions" the UAV's dynamic adaptability in complex environments to the offline planning stage, so that the final generated global path is inherently robust against sudden changes in terrain, thereby ensuring that even if the UAV encounters unexpected geographical obstacles during actual inspection missions, its basic route can still maintain near-optimal performance.

[0032] Furthermore, before inputting the set of path points into the pre-generated path evaluation fitness function, the method further includes:

[0033] Assume that the set of track points is {P0, P1, P2...P n ,P n+1}P0 and P n+1 They represent the starting point and the last inspection point of the inspection drone respectively; P1, P2...P n are the various path nodes during the flight process; let the coordinates of the take-off point and the last inspection point be expressed as P0 = (x0, y0, z0), P n+1 =(x n+1 ,y n+1 ,z n+1 ), use P i =(x i ,y i ,z i) to describe each intermediate node, i = 1, 2, 3...n, the total range of the inspection drone is l al is the total flight range of the inspection drone;

[0034] The constraint cost function of the pitch angle and yaw angle of the inspection drone is as follows:

[0035]

[0036] Among them, K ψ K θ are the coefficients of the pitch angle constraint and the yaw angle constraint, ψ max is the maximum pitch angle, θ max is the maximum yaw angle, C ψi 、C θi are the cost functions of the yaw angle and pitch angle of the i-th track point respectively. When the constraint cost value is 0, it means that the flight path at this time meets the UAV dynamic performance constraints;

[0037] The flight distance constraint cost function is defined as follows:

[0038]

[0039] Among them, L max is the maximum range, C L is the flight distance constraint cost function, K L is the range constraint cost coefficient. When the constraint cost value is 0, it means that the flight range at this time is less than the maximum range.

[0040] The terrain constraint cost function is as follows:

[0041]

[0042] Among them, C hi is the terrain constraint cost function, Z(x i ,y i ) is the altitude value of the i-th track point, Z map (x i ,y i ) is the height value in the three-dimensional environment model under the i-th track point, K h is the cost function value of the terrain, and H is the safe height from the ground;

[0043] The cost function of the no-fly zone in the inspection drone path planning is:

[0044]

[0045] Among them, K R is the flight cost coefficient, C Rijis the constraint cost function of the i-th track point in the j-th no-fly zone, assuming that the projection coordinates of the center of the j-th no-fly zone are The radius is R j , the track path corresponding to track point i The costs associated with The shortest distance d i related;

[0046] The total cost function C of the inspection drone al as follows:

[0047]

[0048] k1+k2+k3+k4+k5=1;

[0049] Among them, n is the number of track points, m is the number of restricted airspace, k1 k2k3 k4k5 are the weights of each cost respectively;

[0050] The fitness function f is as follows:

[0051] f=C al +L al ;

[0052] Among them, C al Greater than L al , to ensure that the flight path of the inspection drone can meet all constraints.

[0053] It's important to note that when drones conduct long-distance, precision inspections in transmission corridors with dramatic three-dimensional terrain and scattered no-fly zones (such as multi-circuit lines crossing canyons), simply pursuing the shortest route or lowest elevation can lead to multiple risks. For example, while flying too close to a mountain may reduce range, it can also lead to loss of control due to excessive pitch angles caused by unexpected air currents. Alternatively, choosing a low-lying route to avoid steep slopes could lead to inadvertent entry into a wildlife sanctuary. Conventional single-cost function optimization often constrains only a single risk dimension. This method simultaneously embeds four physical constraints: range, attitude angle, terrain clearance, and no-fly zones. This allows each track point to undergo a three-dimensional safety assessment. The attitude angle cost function directly anchors the drone's body dynamic limits (e.g., large yaw angles can easily cause camera stabilization to fail). The terrain constraint function enforces a safe buffer between the drone and the ground (to prevent collisions with loose rock or treetops). The no-fly zone function establishes the sensitivity of electronic fences (to avoid legal risks in military zones or ecological reserves). Ultimately, a weighted, global cost function quantifies these conflicting metrics. For example, when planning a route across a canyon, the algorithm doesn't simply choose the shortest straight line (which could result in exceeding the pitch angle limit) nor does it take an overly conservative detour (increasing range risk). Instead, it seeks a balance point that maintains a stable attitude while leaving sufficient ground clearance, while also sticking close to the edge of the no-fly zone but not crossing it (such as flying along the curve of the canyon wall at a safe altitude). This internalized multi-objective, strongly constrained design ensures that the output inspection path pre-digests the main operational risks before actual execution, essentially creating an air channel for drones in complex geographical and regulatory environments that combines physical safety, control stability, and regulatory compliance.

[0054] Furthermore, the Levy flight strategy is performed on each ant individual in the ant population, and based on the influence of the position of the ant lion individual randomly selected in a preset manner and the position of the ant lion to be selected, the path point coordinates of the ant individual in space are updated, including:

[0055] The flight step length L of each ant in the ant population is determined by the following formula: v :

[0056]

[0057] Among them, u and v satisfy the normal distribution. u and v are vectors randomly generated from standard normal distribution, σ u σ v The definition is as follows:

[0058]

[0059] σ v =1;

[0060] Among them, β is [0,2], Γ is the standard Gamma function;

[0061] Based on the influence of the position of the ant lion individual randomly selected in a preset manner and the position of the ant lion to be selected, and combined with the following formula, the coordinates of the path point of the ant individual in space are updated:

[0062] X(t)=[0,cumsum(L v1 ),cumsum(L v2 ),...,cumsum(L vn )];

[0063] Among them, X(t) is the coordinate of the path point of the ant individual in space, cumsum(L vi ) is the accumulation of the positions of the previous i Levy flights.

[0064] Furthermore, the randomly selected ant lion individuals and the ant lions to be selected are processed by the elite pool mechanism and the Gauss-Cauchy hybrid mutation method to obtain processing results, including:

[0065] A preset proportion of population individuals in each iteration is placed into the elite pool, and the elite pool is modeled as follows:

[0066]

[0067] Where Antlion is the sorted antlion individual, t is the current iteration number, and the position information of the antlion individuals is disturbed by randomly selecting antlions from the elite pool and the elite antlions through the Gauss-Cauchy hybrid mutation method to select antlions to prevent the elite antlions from falling into the local optimum.

[0068] According to the above formula, the Gauss-Cauchy hybrid mutation method is applied to the selected ant lion individuals:

[0069]

[0070] Among them, T is the maximum number of iterations, t is the current number of iterations, Antlion is the antlion that undergoes mutation, K GC is the coefficient of variation.

[0071] It's important to note that when power lines traverse valleys and mountain ranges, the undulating terrain creates numerous local optimal traps (for example, a low-altitude saddle path between adjacent peaks may appear optimal, but in reality, observations are lost due to ridge obstruction). Conventional optimization algorithms are easily trapped by these traps and unable to escape. This method first uses elite pool modeling to consolidate the optimal pathpoints from each generation (retaining samples of paths that successfully traverse mountain ridges), continuously infusing the population with effective historical experience. Furthermore, a Gaussian-Cauchy hybrid mutation constructs a differentiated perturbation mechanism using a dual probability distribution. Gaussian mutation fine-tunes the ant lion's position using its bell-shaped distribution (for example, adjusting the elevation of pathpoints to maintain a safe gap when the drone is navigating the mountainside), while Cauchy mutation, with its heavy-tailed nature, grants the ant lion the ability to leap large distances (for example, launching a pathpoint trapped in a valley directly across a peak line to the adjacent slope). This mechanism, in which the intensity of disturbances adapts to the number of iterations (dynamic attenuation of the coefficient of variation), encourages Cauchy mutation in the early stages to push path point clusters beyond terrain isolation (e.g., jumping from the bottom of a canyon to a ridge observation point), and later relies on Gaussian mutation for fine convergence around elite paths (e.g., optimizing the observation angle of hovering sideways on a cable tower). The resulting path evolution paradigm enables the algorithm to handle highly complex power geography environments by inheriting the obstacle avoidance logic of historical optimal paths (elite pool anti-degradation) while also actively breaking through local spatial constraints (hybrid mutation breaking barriers), thereby generating basic routes for drones with terrain traversal intelligence. This completely avoids the risk of missed inspections or aborted flights caused by path point clusters being trapped in local low-risk areas.

[0072] One or more embodiments of this specification provide a device for generating a power inspection path, the device comprising:

[0073] A map determination unit loads offline map information covered by the power inspection according to the operation flight mission of the power inspection UAV to obtain power inspection marked map information, wherein the operation flight mission includes the take-off point location and the locations of each power inspection task point;

[0074] a flight path determining unit, configured to determine a power inspection flight path based on the power inspection mark map information, wherein the power inspection flight path is a path that takes off from the take-off point and passes through each of the power inspection task points;

[0075] Inspection space construction unit, which constructs a local inspection space between the current position of the inspection drone and the local target point;

[0076] An obstacle detection unit, configured to detect sudden obstacle information in the local inspection space using sensors carried by the inspection drone;

[0077] An adjustment strategy determining unit, which obtains a real-time local path adjustment strategy based on the sudden obstacle information;

[0078] A path adjustment unit performs real-time path adjustment on the power inspection flight path based on the real-time local path adjustment strategy.

[0079] One or more embodiments of this specification provide a power inspection path generation device, including:

[0080] at least one processor; and,

[0081] a memory communicatively connected to the at least one processor; wherein,

[0082] The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:

[0083] Loading offline map information covered by the power inspection according to the operational flight mission of the power inspection UAV to obtain power inspection marked map information, wherein the operational flight mission includes the take-off point location and the locations of each power inspection task point;

[0084] Determine a power inspection flight path based on the power inspection mark map information, where the power inspection flight path takes off from the take-off point and passes through each of the power inspection task points;

[0085] Construct a local inspection space between the current position of the inspection drone and the local target point;

[0086] Detecting sudden obstacle information in the local inspection space by using sensors carried by the inspection drone;

[0087] Based on the sudden obstacle information, a real-time local path adjustment strategy is obtained;

[0088] The power inspection flight path is adjusted in real time based on the real-time local path adjustment strategy.

[0089] At least one of the above technical solutions adopted in the embodiments of this specification can achieve the following beneficial effects:

[0090] Because unpredictable and sudden obstacles (such as wild animals or severe weather changes) often appear in power inspection areas (such as high-voltage transmission lines or remote mountainous areas), these obstacles can interrupt the planned flight path, increase the risk of collision, or cause the inspection mission to fail. This method can instantly identify these interference factors and generate adjustment strategies through the construction of local inspection space and real-time sensor detection. Through this dynamic adjustment mechanism, the inspection path can bypass obstacles, maintaining the continuous flight and stable observation of the drone, thereby directly improving the safety and execution efficiency of the mission. This avoids the need for manual intervention or mission termination, ensures the reliability of the inspection process and the integrity of data collection, and ultimately achieves more efficient autonomous inspections in complex and changing environments, reducing overall operational risks and optimizing resource utilization. BRIEF DESCRIPTION OF THE DRAWINGS

[0091] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the following briefly introduces the drawings required for the embodiments or the description of the prior art. Obviously, the drawings described below are only some of the embodiments described in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without inventive work. In the drawings:

[0092] Figure 1 A schematic flow chart of a method for generating a power inspection path according to one or more embodiments of this specification;

[0093] Figure 2 A Gaussian-Cauchy probability comparison diagram provided for one or more embodiments of this specification;

[0094] Figure 3 A schematic diagram of the structure of a power inspection path generation device provided in one or more embodiments of this specification;

[0095] Figure 4 A schematic diagram of the structure of a power inspection path generation device provided in one or more embodiments of this specification. DETAILED DESCRIPTION

[0096] The embodiments of this specification provide a method, device, and equipment for generating a power inspection path.

[0097] To help those skilled in the art better understand the technical solutions in this specification, the following will provide a clear and complete description of the technical solutions in the embodiments of this specification, in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of this specification, not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this specification without creative work should fall within the scope of protection of this specification.

[0098] Figure 1 This is a flow chart of a method for generating a power inspection path according to one or more embodiments of this specification. The flow can be executed by a power inspection path generation system. Certain input parameters or intermediate results in the flow can be manually adjusted to help improve accuracy.

[0099] The method steps of the embodiment of this specification are as follows:

[0100] S101, loading offline map information covered by the power inspection according to the operation flight mission of the power inspection UAV to obtain power inspection mark map information, wherein the operation flight mission includes the take-off point location and the locations of each power inspection task point.

[0101] In the embodiments of this specification, offline map data (such as satellite images or digital elevation models) can be loaded from a local database or cloud storage based on the preset operational flight mission of the power inspection drone (including the take-off point and the location of each mission point), and power equipment marks (such as transmission towers and insulator locations) can be superimposed to generate a spatially marked map that integrates geographic information and inspection targets, which serves as the basic input for subsequent path planning.

[0102] S102 , determining a power inspection flight path based on the power inspection mark map information, where the power inspection flight path takes off from the take-off point and passes through the power inspection task point locations.

[0103] In the embodiment of this specification, based on the marked map information output by S101, an optimization algorithm (such as the improved ant lion algorithm) can be used to automatically calculate the global flight path. The path must strictly connect the take-off point and all inspection task points, and ensure that the track is continuous and flyable through smoothing processing to form a complete initial inspection route, providing the drone with a structured navigation framework from the starting point to the end point.

[0104] S103, constructing a local inspection space between the current position of the inspection drone and the local target point.

[0105] In the embodiment of this specification, when the UAV flies to any section along the global path generated by S102, a local three-dimensional scanning space (the range is constrained by the preset safety radius and flight altitude) is dynamically constructed in real time based on its current GPS positioning and the coordinates of the next task point, and the space is defined as a temporary operation area that needs to be monitored first to focus on subsequent obstacle detection.

[0106] S104: Detecting information about sudden obstacles in the local inspection space by using sensors carried by the inspection drone.

[0107] In the embodiment of this specification, within the local inspection space defined in S103, multi-source sensors (such as lidar or visual cameras) carried by the drone can be used to perform real-time environmental scanning, identify the location and outline of sudden obstacles (such as flying birds, temporary construction equipment), and output obstacle attribute data, providing real-time threat input for dynamic path adjustment.

[0108] S105: Obtain a real-time local path adjustment strategy based on the sudden obstacle information.

[0109] In the embodiment of this specification, based on the sudden obstacle information detected by S104, combined with the local space boundary and the UAV maneuvering constraints (such as turning radius), an obstacle avoidance algorithm (such as a fast random expanding tree) can be called to generate a detour or redirection strategy to ensure that the new path avoids obstacles and maintains the smoothness of the route, and output a local track correction instruction that can be directly executed.

[0110] S106 , performing real-time path adjustment on the power inspection flight path based on the real-time local path adjustment strategy.

[0111] In the embodiments of this specification, the real-time local path adjustment strategy of S105 can be seamlessly integrated with the global flight path of S102, and the route can be corrected online (such as inserting detour points or deleting conflict segments) by dynamically updating the waypoint sequence, so that the UAV can respond to local changes in real time under the global mission framework, ensuring the continuity and safety of the inspection mission.

[0112] It should be noted that unpredictable and sudden obstacles (such as wild animals or severe weather changes) often appear in power inspection areas (such as high-voltage transmission lines or remote mountainous areas). These obstacles will interrupt the planned flight path, increase the risk of collision or cause the inspection mission to fail. This method can instantly identify these interference factors and generate adjustment strategies through the construction of local inspection space and real-time sensor detection. Through this dynamic adjustment mechanism, the inspection path can bypass obstacles, maintain the continuous flight and stable observation of the drone, and thus directly improve the safety and execution efficiency of the mission. This avoids the need for manual intervention or mission termination, ensures the reliability of the inspection process and the integrity of data collection, and ultimately achieves more efficient autonomous inspections in complex and changing environments, reducing overall operational risks and optimizing resource utilization.

[0113] Furthermore, when determining the power inspection flight path based on the power inspection mark map information, the parameter information of the ant lion algorithm can be set, and the parameter information includes the number of ant populations, the number of ant lion populations, the number of iterations and the dimension, and the dimension includes the three-dimensional coordinates of multiple path points generated by individual ants and individual ant lions in space; the parameter information of the ant lion algorithm is initialized through a pre-set chaotic mapping to generate path points of individual ants and individual ant lions in space; the path points are smoothed to generate a set of path points; and the power inspection flight path is generated based on the set of path points.

[0114] It should be noted that for the above content, the core parameters of the antlion optimization algorithm can be preset. For example, the population size of ants and antlions (e.g., 50 each), the total number of iterations (e.g., 100), and the three-dimensional spatial dimensions of the path points can be defined (each individual represents a path consisting of three-dimensional coordinate points). Preset rules such as Sine chaotic mapping are used to initialize the spatial positions of the ant and antlion populations, generating a series of random but highly traversable three-dimensional path points (covering the inspection area) for each individual, thus avoiding the uneven distribution problem caused by traditional random initialization. All three-dimensional path points generated by chaotic initialization are then subjected to trajectory smoothing. That is, adjacent discrete points can be connected using cubic spline interpolation technology to generate a set of path points with continuous curvature transitions, eliminating sudden turns in the trajectory. Based on the smoothed path point set, the starting point, intermediate points, and end point are connected in sequence, and a three-dimensional continuous flight path that conforms to the drone's dynamic performance is directly output as the basic route for the global inspection mission.

[0115] It should be noted that in complex three-dimensional scenarios such as high-voltage transmission line inspections, path planning must balance geographical constraints (e.g., avoiding mountains and vegetation), equipment accessibility (e.g., proximity to insulator strings), and flight efficiency. Traditional planning methods are susceptible to sudden changes in terrain or densely packed equipment layouts, resulting in risks such as sharp turns, ineffective circling, and proximity to obstacles. This method utilizes chaotic mapping to optimize the initialization process of the ant lion algorithm. Essentially, it uses nonlinear mapping to improve the spatial traversability and uniformity of the population's initial positions. Combined with path point smoothing, the distribution of the generated initial path points in three-dimensional space is more adapted to the layout characteristics of power facilities (e.g., natural extension along the line direction and smooth transitions). This not only avoids the problem of random initialization easily falling into local inferior solutions, but also forms a continuous and smooth spatial curve. This basic path characteristic enables subsequent UAVs to more naturally adapt to the spatial distribution characteristics of inspection targets during inspections, reducing the risk of control instability, equipment observation angle deviation, and excessive maneuvering time loss caused by sharp path fluctuations or frequent sharp turns, thus laying the foundation for efficient and stable autonomous inspections by UAVs in complex geographical environments.

[0116] Furthermore, the traditional ALO algorithm uses random generation to initialize the ant and ant lion populations. This method reduces the diversity of the initial population, resulting in large differences in the generated ant lion and ant populations. The environmental information of the search space cannot be fully utilized. The relatively poor ant and ant lion individuals in the population will reduce the accuracy of the optimization target solution, making the quality of the optimization result heavily dependent on the generated initial population, affecting the robustness of the algorithm. Chaotic mapping is a method for studying chaotic behavior in dynamic systems. Common chaotic mappings include Logistic mapping and Henon mapping. The original random methods of the sequences generated by these mappings have certain similarities, but there are also some differences. Chaotic mapping usually has the following characteristics: (1) Pseudo-randomness, the trajectory of the chaotic phenomenon is sensitive to the initial conditions; (2) Regularity, each generation can traverse the entire population space. (3) Traversability, the trajectory of the chaotic mapping can traverse all positions within a certain range without repetition.

[0117] To this end, this application uses an improved Sine chaotic map to generate the initial ant and ant lion populations to ensure that the initial population has good spatial traversability in the solution space and is evenly distributed in the phase space. The improved Sine chaotic map is shown in formula (1). In the planning problem, the initialized ant lion population should include the location information of the path points through the following formula (2).

[0118] When the parameter information of the ant lion algorithm is initialized by the pre-set chaotic map and the path points of the ant individuals and ant lion individuals in space are generated, the Initialize the parameter information of the ant lion algorithm and generate the path points of ant individuals and ant lion individuals in space

[0119]

[0120] Among them, d i With e i are all random numbers (0,1), u is the control parameter, w i+1 is the iterative chaotic sequence value, (x n,m ,y n,m ,z n,m ) is the coordinate of the nth ant lion individual, the mth path point along the X, Y, and Z axes, n is the maximum population size, and m is the number of path points included in the set ant lion individual.

[0121] Furthermore, when the path points are smoothed and a path point set is generated, the path points can be smoothed using a cubic spline interpolation method to associate adjacent path points in the flight path of the power inspection drone. The multiple interpolation points and the starting and ending points obtained constitute a track point set, which are connected to obtain the flight path set of the power inspection drone.

[0122] It's important to note that in scenarios with densely distributed power equipment (such as substations or transmission line corridors), while the basic path points generated by the Ant Lion algorithm avoid major obstacles, the spatial jumps between discrete points mean that connecting them directly with straight lines will result in a zigzag path with sharp turns. This path forces the drone to frequently accelerate and decelerate, making sharp turns, which not only increases motor energy consumption and control burdens, but also causes the camera to lose precise focus on key inspection targets (such as cracks in insulator strings or conductor connection points) due to severe shaking. Cubic spline interpolation, on the other hand, uses mathematical constraints to automatically generate a continuous, smooth curvature transition curve between adjacent path points. This curve possesses the core property of second-order derivative continuity, meaning that the drone's steering angular velocity and acceleration remain constant as it flies along this curve. This results in a dual optimization: on the flight side, by eliminating the risk of sudden changes in attitude, the flight control system can maintain a stable flight attitude, avoiding sudden turns or positioning drift. On the observation side, the onboard detection equipment can maintain its optical axis aligned with the target while flying at a constant speed and low jitter, ensuring effective image acquisition time and image clarity at each inspection point. This flight smoothness, derived from mathematical continuity, essentially lays an "invisible track" for drones in complex space operations, enabling them to balance safety and precision when performing detailed inspection tasks, ultimately achieving high-quality autonomous inspections with full coverage of equipment.

[0123] Furthermore, when generating the power inspection flight path based on the path point set, the path point set can be input into a pre-generated path evaluation fitness function to obtain the fitness value of each path point, and the minimum fitness value is marked as the candidate ant lion; in the current number of iterations, the Levy flight strategy is performed on each ant individual in the ant population, and based on the influence of the position of the ant lion individual randomly selected in a preset manner and the position of the candidate ant lion, the path point coordinates of the ant individual in space are updated; the randomly selected ant lion individual and the candidate ant lion are processed by the elite pool mechanism and the Gauss-Cauchy mixed mutation method to obtain a processing result; based on the processing result, the fitness value of the ant individual and the ant lion individual after the wandering is calculated; the ant population and the ant lion population are sorted based on the fitness value, and the individual with the smallest fitness value in the historical iteration is marked as the elite ant lion.

[0124] It should be noted that in power line scenarios with interlaced mountainous areas and canyons, the drone path needs to avoid obstacles in three dimensions (such as avoiding protruding ridges, high-voltage towers, and meteorological equipment), while also meeting the viewing angle adjustment requirements for fine inspections (such as controlling the close observation angle of insulator strings). Traditional optimization algorithms are prone to falling into suboptimal paths due to dimensionality explosion or local terrain traps (such as repeatedly circling the same valley or excessively raising the route, resulting in observation failure). This method first uses the long and short jump characteristics of Levy flight (random step length based on power-law distribution) to enable individual ants to have both large-scale exploration and small-step fine-tuning capabilities when following the movement of the ant lion. This means that when a cluster of pathpoints falls into a local depression (such as surrounded by a ring of mountains), long jump steps can directly cross the terrain barrier to explore new areas, avoiding the group's repeated consumption of computing power in invalid areas. Simultaneously, the elite pool mechanism continuously retains the optimal fitness solution (candidate antlions) of each generation and applies perturbations of varying intensities to elite individuals through Gaussian-Cauchy hybrid mutation. Gaussian mutation fine-tunes coordinate points to optimize subtle flight postures (such as translating pathpoints to aim the camera at insulator hardware), while Cauchy mutation confers the ability to make large coordinate jumps (such as jumping directly from one side of a valley to the other to avoid an entire thunderstorm area). This strategy of "gradual optimization and radical breakthrough" coexists, inheriting the genes of historical high-quality paths while actively injecting diversity. Finally, using the path evaluation function (which integrates weights such as flight distance, safety margin, and observation angle) as the screening criterion, the elite antlion individuals formed after multiple rounds of iteration naturally represent a path set that naturally combines the comprehensive advantages of obstacle avoidance safety, movement efficiency, and observation integrity. Essentially, this algorithm structure "pre-positions" the UAV's dynamic adaptability in complex environments to the offline planning stage, so that the final generated global path is inherently robust against sudden changes in terrain, thereby ensuring that even if the UAV encounters unexpected geographical obstacles during actual inspection missions, its basic route can still maintain near-optimal performance.

[0125] Furthermore, before inputting the path point set into the pre-generated path evaluation fitness function, the set of track points can be set to be {P0, P1, P2...P n ,P n+1}P0 and P n+1 They represent the starting point and the last inspection point of the inspection drone respectively; P1, P2...P n are the various path nodes during the flight process; let the coordinates of the take-off point and the last inspection point be expressed as P0 = (x0, y0, z0), P n+1 =(x n+1 ,y n+1 ,z n+1 ), use P i =(x i ,y i ,z i ) to describe each intermediate node, i = 1, 2, 3...n, the total range of the inspection drone is L al is the total flight range of the inspection drone;

[0126] In order to ensure the safety and feasibility of inspection drones when performing inspection tasks, the constraints during simulation need to be more in line with actual needs. Therefore, in addition to considering the safety of flight missions, this application also needs to consider its own dynamic constraints in the planning problem of inspection drones.

[0127] Inspection drones are subject to physical constraints during their climb and turn operations. If the climb or turn angle is too large, the drone can easily stall and crash, seriously impacting flight safety. The constraint cost functions for the pitch and yaw angles of inspection drones are as follows:

[0128]

[0129] Among them, K ψ K θ are the coefficients of the pitch angle constraint and the yaw angle constraint, ψ max is the maximum pitch angle, θ max is the maximum yaw angle, C ψi 、C θi are the cost functions of the yaw angle and pitch angle of the i-th track point respectively. When the constraint cost value is 0, it means that the flight path at this time meets the UAV dynamic performance constraints;

[0130] Due to the limitation of battery capacity of inspection drones, the performance parameters of drones include their maximum flight range. As a result, their total flight range during autonomous inspections should be less than the maximum flight range of the drone. The flight distance constraint cost function is defined as follows:

[0131]

[0132] Among them, L max is the maximum range, C L is the flight distance constraint cost function, K L is the range constraint cost coefficient. When the constraint cost value is 0, it means that the flight range at this time is less than the maximum range.

[0133] The fitness function of an inspection drone's path is an important evaluation indicator for determining whether the flight path meets the mission requirements. The design of the drone's path fitness function requires comprehensive consideration of multiple factors to ensure that the generated flight path can meet the inspection mission while maintaining high safety and flight efficiency. Based on the analysis of the inspection mission requirements, the fitness function should include two aspects: the cost function and the path length. The cost function mainly includes:

[0134] Flight distance cost function C LFlight length is an important indicator for evaluating path planning solutions. A shorter path length usually indicates a more efficient path planning solution.

[0135] Flight obstacle cost function C Rij and C hi The flight obstacle cost function should consider the safety of the flight path to avoid danger caused by the inspection drone's flight path in the no-fly zone or collision with the terrain during flight. The terrain constraint cost function is as follows:

[0136]

[0137] Among them, C hi is the terrain constraint cost function, Z(x i ,y i ) is the altitude value of the i-th track point, Z map (x i ,y i ) is the height value in the three-dimensional environment model under the i-th track point, K h is the cost function value of the terrain, and H is the safe height from the ground;

[0138] The cost function of the no-fly zone in the inspection drone path planning is:

[0139]

[0140] Among them, K R is the flight cost coefficient, C Rij is the constraint cost function of the i-th track point in the j-th no-fly zone, assuming that the projection coordinates of the center of the j-th no-fly zone are The radius is R j , the track path corresponding to track point i The costs associated with The shortest distance d i related;

[0141] Inspection drone's own cost function C ψi and C θi Due to the limitations of the drone's physical properties and maximum flight distance, the flight path should be within certain limits to avoid excessive climbing and turning angles or the total flight path exceeding the maximum flight path to ensure flight stability.

[0142] Finally, the total cost function C of the inspection drone is obtained by weighting the costs of the terrain constraints, no-fly zone constraints and the drone's own constraints. al , as follows:

[0143]

[0144] k1+k2+k3+k4+k5=1;

[0145] Where n is the number of track points, m is the number of restricted airspace, k1 k2k3 k4k5 are the weights of each cost, k1 = 0.2, k2 = 0.2, k3 = 0.2, k4 = 0.3, k5 = 0.3, and the cost function value of all constraints is a larger number;

[0146] The fitness function f is as follows:

[0147] f=C al +L al ;

[0148] Among them, C al Can be much larger than L al , to ensure that the flight path of the inspection drone can meet all constraints.

[0149] It's important to note that when drones conduct long-distance, precision inspections in transmission corridors with dramatic three-dimensional terrain and scattered no-fly zones (such as multi-circuit lines crossing canyons), simply pursuing the shortest route or lowest elevation can lead to multiple risks. For example, while flying too close to a mountain may reduce range, it can also lead to loss of control due to excessive pitch angles caused by unexpected air currents. Alternatively, choosing a low-lying route to avoid steep slopes could lead to inadvertent entry into a wildlife sanctuary. Conventional single-cost function optimization often constrains only a single risk dimension. This method simultaneously embeds four physical constraints: range, attitude angle, terrain clearance, and no-fly zones. This allows each track point to undergo a three-dimensional safety assessment. The attitude angle cost function directly anchors the drone's body dynamic limits (e.g., large yaw angles can easily cause camera stabilization to fail). The terrain constraint function enforces a safe buffer between the drone and the ground (to prevent collisions with loose rock or treetops). The no-fly zone function establishes the sensitivity of electronic fences (to avoid legal risks in military zones or ecological reserves). Ultimately, a weighted, global cost function quantifies these conflicting metrics. For example, when planning a route across a canyon, the algorithm doesn't simply choose the shortest straight line (which could result in exceeding the pitch angle limit) nor does it take an overly conservative detour (increasing range risk). Instead, it seeks a balance point that maintains a stable attitude while leaving sufficient ground clearance, while also sticking close to the edge of the no-fly zone but not crossing it (such as flying along the curve of the canyon wall at a safe altitude). This internalized multi-objective, strongly constrained design ensures that the output inspection path pre-digests the main operational risks before actual execution, essentially creating an air channel for drones in complex geographical and regulatory environments that combines physical safety, control stability, and regulatory compliance.

[0150] Furthermore, Lévy flight, a key paradigm for non-Gaussian random processes, is mathematically characterized by a heavy-tailed power-law distribution of displacements. Proposed by renowned French mathematician Paul Lévy, the algorithm's core lies in the significant alternation of long and short steps during its movement. Compared to the continuous, small-amplitude migrations of traditional Brownian motion, the Lévy process, through the tailing property of the probability density function, exhibits intermittent bursts of displacement during system evolution. This unique motion pattern exhibits both ergodic and random characteristics, belonging to a Markov process and a random walk model. The algorithm's flight trajectory in phase space appears as an alternating combination of high-frequency, small-step jumps (clustered together) and low-frequency, large-step jumps. Furthermore, it exhibits power-law asymptotic behavior and a "heavy-tailed" distribution, providing a mathematical foundation for optimal search strategies in complex environments. Incorporating this into the ALO algorithm allows for full exploration of the solution domain. The early long-step jumps enrich the population's diversity, significantly reducing the risk of ant lions becoming trapped in local optima. In the later stage, the convergence of the algorithm is accelerated based on its heavy-tail characteristics, which can improve the quality of the algorithm solution.

[0151] The Levy flight strategy is performed on each ant individual in the ant population, and based on the influence of the position of the ant lion individual randomly selected in a preset manner and the position of the ant lion to be selected, when updating the path point coordinates of the ant individual in the space, the flight step length L of each ant individual in the ant population can be determined by the following formula: v :

[0152]

[0153] Among them, u and v satisfy the normal distribution. u and v are vectors randomly generated from standard normal distribution, σ u σ v The definition is as follows:

[0154]

[0155] σ v =1;

[0156] Among them, β is [0,2], Γ is the standard Gamma function;

[0157] Based on the influence of the position of the ant lion individual randomly selected in a preset manner and the position of the ant lion to be selected, and combined with the following formula, the coordinates of the path point of the ant individual in space are updated:

[0158] X(t)=[0,cumsum(L v1 ),cumsum(L v2 ),...,cumsum(L vn )];

[0159] Among them, X(t) is the coordinate of the path point of the ant individual in space, cumsum(L vi ) is the accumulation of the positions of the previous i Levy flights.

[0160] Furthermore, in the original ALO algorithm, the elite antlion is defined as the individual with the best fitness during the iteration process. This fitness is primarily determined by the random walk of the ant population during the iterations, and the ALO algorithm itself does not actively perturb the elite antlions. However, the ALO algorithm's iterations show that when the algorithm is stuck in a local optimum, the elite antlions are also trapped in that local optimum. The updates of individual ants rely on the guidance of the current elite antlion and the antlions selected by the roulette wheel, causing the updates of individual ant positions to remain within the local optimum and prevent them from escaping. To address this issue, the concept of an elite pool is introduced.

[0161] The randomly selected ant lion individuals and the ant lions to be selected are processed by the elite pool mechanism and the Gauss-Cauchy hybrid mutation method. When the processing results are obtained, a preset proportion of population individuals in each iteration can be put into the elite pool. The preset proportion can be 10%. The elite pool is modeled as follows:

[0162]

[0163] Among them, Antlion is the sorted antlion individual, t is the current iteration number, and the antlion individuals and elite antlions randomly selected from the elite pool are perturbed by the Gauss-Cauchy hybrid mutation method to select antlion individuals to prevent the elite antlion individuals from falling into the local optimum and affecting the overall optimization ability of the algorithm.

[0164] According to the above formula, the Gauss-Cauchy hybrid mutation method is applied to the selected ant lion individuals:

[0165]

[0166] Among them, T is the maximum number of iterations, t is the current number of iterations, Antlion is the antlion that undergoes mutation, K GC is the coefficient of variation.

[0167] See also Figure 2The Gaussian-Cauchy probability comparison diagram shown in Figure 2 shows that the algorithm's dynamic weight adjustment mechanism exhibits significant phased nature. In the initial iterations, when the t value is small, the Cauchy mutation dominates. The long-range jumps generated by its distribution characteristics effectively expand the phase space exploration range and suppress the algorithm's premature convergence. As the number of iterations increases, the Gaussian mutation weight gradually becomes dominant. At this point, the algorithm focuses on the search accuracy of the local optimal space, achieving progressive optimization of the solution vector through the high-density probability characteristics of the Gaussian distribution. Through the synergistic effect of this hybrid mutation strategy, both sufficient exploration of the global search space and high-precision convergence to the optimal position space are guaranteed.

[0168] It's important to note that when power lines traverse valleys and mountain ranges, the undulating terrain creates numerous local optimal traps (for example, a low-altitude saddle path between adjacent peaks may appear optimal, but in reality, observations are lost due to ridge obstruction). Conventional optimization algorithms are easily trapped by these traps and unable to escape. This method first uses elite pool modeling to consolidate the optimal pathpoints from each generation (retaining samples of paths that successfully traverse mountain ridges), continuously infusing the population with effective historical experience. Furthermore, a Gaussian-Cauchy hybrid mutation constructs a differentiated perturbation mechanism using a dual probability distribution. Gaussian mutation fine-tunes the ant lion's position using its bell-shaped distribution (for example, adjusting the elevation of pathpoints to maintain a safe gap when the drone is navigating the mountainside), while Cauchy mutation, with its heavy-tailed nature, grants the ant lion the ability to leap large distances (for example, launching a pathpoint trapped in a valley directly across a peak line to the adjacent slope). This mechanism, in which the intensity of disturbances adapts to the number of iterations (dynamic attenuation of the coefficient of variation), encourages Cauchy mutation in the early stages to push path point clusters beyond terrain isolation (e.g., jumping from the bottom of a canyon to a ridge observation point), and later relies on Gaussian mutation for fine convergence around elite paths (e.g., optimizing the observation angle of hovering sideways on a cable tower). The resulting path evolution paradigm enables the algorithm to handle highly complex power geography environments by inheriting the obstacle avoidance logic of historical optimal paths (elite pool anti-degradation) while also actively breaking through local spatial constraints (hybrid mutation breaking barriers), thereby generating basic routes for drones with terrain traversal intelligence. This completely avoids the risk of missed inspections or aborted flights caused by path point clusters being trapped in local low-risk areas.

[0169] Figure 3 This is a structural diagram of a power inspection path generation device provided for one or more embodiments of this specification, wherein the device includes: a map determination unit 301, a flight path determination unit 302, an inspection space construction unit 303, an obstacle detection unit 304, an adjustment strategy determination unit 305 and a path adjustment unit 306.

[0170] The map determination unit 301 loads offline map information covered by the power inspection according to the operation flight mission of the power inspection UAV to obtain power inspection marked map information. The operation flight mission includes the take-off point location and the locations of each power inspection task point;

[0171] A flight path determining unit 302 is configured to determine a power inspection flight path based on the power inspection mark map information, wherein the power inspection flight path is a path that takes off from the take-off point and passes through each of the power inspection task points.

[0172] The inspection space construction unit 303 constructs a local inspection space between the current position of the inspection drone and the local target point;

[0173] The obstacle detection unit 304 detects sudden obstacle information in the local inspection space through the sensors carried by the inspection drone;

[0174] An adjustment strategy determining unit 305 obtains a real-time local path adjustment strategy based on the sudden obstacle information;

[0175] The path adjustment unit 306 performs real-time path adjustment on the power inspection flight path based on the real-time local path adjustment strategy.

[0176] Figure 4 A schematic structural diagram of a power inspection path generation device provided for one or more embodiments of this specification includes:

[0177] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:

[0178] According to the operational flight mission of the power inspection UAV, the offline map information covered by the power inspection is loaded to obtain the power inspection mark map information, wherein the operational flight mission includes the take-off point location and the locations of each power inspection task point; based on the power inspection mark map information, the power inspection flight path is determined, and the power inspection flight path is a path that takes off from the take-off point location and passes through the locations of each power inspection task point; a local inspection space is constructed between the current position of the inspection UAV and the local target point; sudden obstacle information is detected in the local inspection space by the sensors carried by the inspection UAV; based on the sudden obstacle information, a real-time local path adjustment strategy is obtained; and the power inspection flight path is adjusted in real time based on the real-time local path adjustment strategy.

[0179] The various embodiments in this specification are described in a progressive manner. Similar portions between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from the other embodiments. In particular, the device, apparatus, and non-volatile computer storage medium embodiments are generally similar to the method embodiments, so their descriptions are relatively simplified. For relevant details, refer to the descriptions of the method embodiments.

[0180] The various embodiments in this specification are described in a progressive manner. Similar parts between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences from other embodiments. In particular, the device embodiments are generally similar to the method embodiments, so the description is relatively simple. For relevant parts, refer to the description of the method embodiments.

[0181] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0182] In the embodiments provided in this application, it should be understood that the disclosed devices / network equipment and methods can be implemented in other ways. For example, the device / network equipment embodiments described above are merely illustrative. For example, the division of the modules or units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0183] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0184] In addition, the functional units in the various embodiments of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above units may be implemented in the form of hardware or software.

[0185] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the process in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by the processor, it can implement the steps of the above-mentioned various method embodiments. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0186] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.

Claims

1. A method for generating a power inspection path, characterized in that: The method comprises: Loading offline map information covered by the power inspection according to the operational flight mission of the power inspection UAV to obtain power inspection marked map information, wherein the operational flight mission includes the take-off point location and the locations of each power inspection task point; Determine a power inspection flight path based on the power inspection mark map information, where the power inspection flight path takes off from the take-off point and passes through each of the power inspection task points; Construct a local inspection space between the current position of the inspection drone and the local target point; Detecting sudden obstacle information in the local inspection space by using sensors carried by the inspection drone; Based on the sudden obstacle information, a real-time local path adjustment strategy is obtained; The power inspection flight path is adjusted in real time based on the real-time local path adjustment strategy.

2. The method according to claim 1, characterized in that The determining of the power inspection flight path based on the power inspection mark map information includes: Setting parameter information of the ant lion algorithm, including the number of ant populations, the number of ant lion populations, the number of iterations, and the dimension, wherein the dimension includes the three-dimensional coordinates of multiple path points generated by individual ants and ant lions in space; Initializing parameter information of the ant lion algorithm through a pre-set chaotic map to generate path points of ant individuals and ant lion individuals in space; Smoothing the path points to generate a path point set; A power inspection flight path is generated based on the set of path points.

3. The method according to claim 2, characterized in that Initializing the parameter information of the ant lion algorithm by a preset chaotic map to generate path points of ant individuals and ant lion individuals in space includes: pass Initialize the parameter information of the ant lion algorithm and generate the path points of ant individuals and ant lion individuals in space Among them, d i With e i are all random numbers (0,1), u is the control parameter, w i+1 is the iterative chaotic sequence value, (x n,m ,y n,m ,z n,m ) is the coordinate of the nth ant lion individual, the mth path point along the X, Y, and Z axes, n is the maximum population size, and m is the number of path points included in the set ant lion individual.

4. The method according to claim 2, characterized in that The smoothing process on the path points to generate a path point set includes: The path points are smoothed using the cubic spline interpolation method to associate adjacent path points in the flight path of the power inspection drone. The resulting multiple interpolation points and the starting and ending points constitute a track point set, which are connected to obtain the flight path set of the power inspection drone.

5. The method according to claim 2, characterized in that Generating a power inspection flight path based on the set of path points includes: Input the set of path points into a pre-generated path evaluation fitness function to obtain the fitness value of each path point, and mark the minimum fitness value as the candidate ant lion; In the current iteration, the Levy flight strategy is applied to each ant in the ant population, and the coordinates of the ant's path points in space are updated based on the influence of the position of the ant lion randomly selected in a preset manner and the position of the candidate ant lion. The randomly selected ant lion individual and the candidate ant lion are processed by the elite pool mechanism and the Gauss-Cauchy hybrid mutation method to obtain a processing result; Calculating the fitness values ​​of the individual ants and the individual ant lions after the wandering is completed based on the processing results; The ant population and the ant lion population are sorted based on the fitness value, and the individual with the smallest fitness value in the historical iteration is marked as an elite ant lion.

6. The method according to claim 5, characterized in that Before inputting the set of path points into a pre-generated path evaluation fitness function, the method further includes: Assume that the set of track points is {P0, P1, P2...P n ,P n+1 P0 and P n+1 They represent the starting point and the last inspection point of the inspection drone respectively; P1, P2...P n are the various path nodes during the flight process; let the coordinates of the take-off point and the last inspection point be expressed as P0 = (x0, y0, z0), P n+1 =(x n+1 ,y n+1 ,z n+1 ), use P i =(x i ,y i ,z i ) to describe each intermediate node, i = 1, 2, 3...n, the total range of the inspection drone is L al is the total flight range of the inspection drone; The constraint cost function of the pitch angle and yaw angle of the inspection drone is as follows: Among them, K ψ K θ are the coefficients of the pitch angle constraint and the yaw angle constraint, ψ max is the maximum pitch angle, θ max is the maximum yaw angle, C ψi 、C θi are the cost functions of the yaw angle and pitch angle of the i-th track point respectively. When the constraint cost value is 0, it means that the flight path at this time meets the UAV dynamic performance constraints; The flight distance constraint cost function is defined as follows: Among them, L max is the maximum range, C L is the flight distance constraint cost function, K L is the range constraint cost coefficient. When the constraint cost value is 0, it means that the flight range at this time is less than the maximum range. The terrain constraint cost function is as follows: Among them, C hi is the terrain constraint cost function, Z(x i ,y i ) is the altitude value of the i-th track point, Z map (x i ,y i ) is the height value in the three-dimensional environment model under the i-th track point, K h is the cost function value of the terrain, and H is the safe height from the ground; The cost function of the no-fly zone in the inspection drone path planning is: Among them, K R is the flight cost coefficient, C Rij is the constraint cost function of the i-th track point in the j-th no-fly zone, assuming that the projection coordinate of the center of the j-th no-fly zone is o (xj,yj) , with a radius of R j , the track path corresponding to track point i The cost associated with it (xj,yj) The shortest distance d i related; The total cost function C of the inspection drone al as follows: k1+k2+k3+k4+k5=1; Among them, n is the number of track points, m is the number of restricted airspace, k1 k2k3 k4k5 are the weights of each cost respectively; The fitness function f is as follows: f=C al +L al ; Among them, C al Greater than L al , to ensure that the flight path of the inspection drone can meet all constraints.

7. The method according to claim 5, characterized in that The Levy flight strategy is performed on each ant individual in the ant population, and based on the influence of the position of the ant lion individual randomly selected in a preset manner and the position of the ant lion to be selected, the path point coordinates of the ant individual in space are updated, including: The flight step length L of each ant in the ant population is determined by the following formula: v : Among them, u and v satisfy the normal distribution. u and v are vectors randomly generated from standard normal distribution, σ u σ v The definition is as follows: s v =1; Among them, β is [0,2], Γ is the standard Gamma function; Based on the influence of the position of the ant lion individual randomly selected in a preset manner and the position of the ant lion to be selected, and combined with the following formula, the coordinates of the path point of the ant individual in space are updated: X(t)=[0,cumsum(L v1 ), sum(L v2 ),…,cum(L vn )]: Among them, X(t) is the coordinate of the path point of the ant individual in space, cumsum(L vi ) is the accumulation of the positions of the previous i Levy flights.

8. The method according to claim 5, characterized in that The randomly selected ant lion individuals and the ant lions to be selected are processed by the elite pool mechanism and the Gauss-Cauchy hybrid mutation method to obtain processing results, including: A preset proportion of population individuals in each iteration is placed into the elite pool, and the elite pool is modeled as follows: Where Antlion is the sorted antlion individual, t is the current iteration number, and the position information of the antlion individuals is disturbed by randomly selecting antlions from the elite pool and the elite antlions through the Gauss-Cauchy hybrid mutation method to select antlions to prevent the elite antlions from falling into the local optimum. The formula for applying the Gauss-Cauchy hybrid mutation method on the selected ant lion individuals is: Among them, T is the maximum number of iterations, t is the current number of iterations, Antlion is the antlion that undergoes mutation, K GC is the coefficient of variation.

9. A power inspection path generation device, characterized in that: The device comprises: A map determination unit loads offline map information covered by the power inspection according to the operation flight mission of the power inspection UAV to obtain power inspection marked map information, wherein the operation flight mission includes the take-off point location and the locations of each power inspection task point; a flight path determining unit, configured to determine a power inspection flight path based on the power inspection mark map information, wherein the power inspection flight path is a path that takes off from the take-off point and passes through each of the power inspection task points; Inspection space construction unit, which constructs a local inspection space between the current position of the inspection drone and the local target point; An obstacle detection unit, configured to detect sudden obstacle information in the local inspection space using sensors carried by the inspection drone; An adjustment strategy determining unit, which obtains a real-time local path adjustment strategy based on the sudden obstacle information; A path adjustment unit performs real-time path adjustment on the power inspection flight path based on the real-time local path adjustment strategy.

10. A power inspection path generation device, characterized in that: include: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to: Loading offline map information covered by the power inspection according to the operational flight mission of the power inspection UAV to obtain power inspection marked map information, wherein the operational flight mission includes the take-off point location and the locations of each power inspection task point; Determine a power inspection flight path based on the power inspection mark map information, where the power inspection flight path takes off from the take-off point and passes through each of the power inspection task points; Construct a local inspection space between the current position of the inspection drone and the local target point; Detecting sudden obstacle information in the local inspection space by using sensors carried by the inspection drone; Based on the sudden obstacle information, a real-time local path adjustment strategy is obtained; The power inspection flight path is adjusted in real time based on the real-time local path adjustment strategy.

Images