An environmentally adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants

By constructing forward and reverse directed inspection states, calculating air density and velocity components using a standard atmospheric model, discretizing them into a directed grid graph, eliminating infeasible edges, and generating a minimum energy consumption channel connection matrix, the accuracy and efficiency issues of path planning in UAV inspection of high-altitude photovoltaic power stations are solved, and energy-priority environmental adaptive path planning is realized.

CN123113989APending Publication Date: 2026-11-17四川民族学院
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611514852.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-28
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies for drone inspection of photovoltaic power plants in high-altitude mountainous areas have low accuracy and efficiency in environmentally adaptive multi-channel energy sensing path planning, and cannot effectively cope with the non-uniformity of energy consumption and the differences among multiple feasible channels caused by complex geographical environments.

Method used

By acquiring the digital elevation model of the photovoltaic power station, photovoltaic layout, hard exclusion zone, safety buffer zone, and spatial wind vector field, forward and reverse directional inspection states are constructed. The air density and velocity components are calculated using a standard atmospheric model, discretized into a directed grid graph, infeasible edges are removed, and a graph search is performed to obtain the shortest and lowest energy consumption channels. The connection matrix of the lowest energy consumption channel is generated, and the inspection route is planned under the constraint of available battery energy.

Benefits of technology

It improves the accuracy and efficiency of environmentally adaptive multi-channel energy sensing path planning for UAV inspections of high-altitude photovoltaic power plants, ensuring that UAVs make connection decisions that prioritize energy and have assessable range in complex environments.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123113989A_ABST
    Figure CN123113989A_ABST
Patent Text Reader

Abstract

This invention relates to artificial intelligence technology and discloses an environment-adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants. The method includes: constructing inspection trajectories and speed limits for each photovoltaic row; establishing bidirectional inspection states for each photovoltaic row, and calculating directed inspection energy consumption by combining a high-altitude atmospheric model and local wind field integration; discretizing the free space into a directed raster graph with geometric distance weights and dynamic energy weights, and removing infeasible edges; searching for the shortest distance and lowest energy consumption channels in the same graph, and constructing four sets of directed connection matrices representing the range and energy consumption of the two types of channels; optimizing the photovoltaic row access order in the outer layer, and calling two-state dynamic programming in the inner layer to determine the inspection direction and match the lowest energy consumption channel, outputting a closed inspection route while satisfying battery, obstacle, and performance constraints. This invention can improve the accuracy and efficiency of environment-adaptive multi-channel energy sensing path planning for UAV inspection of high-altitude photovoltaic power plants.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of artificial intelligence, and in particular relates to an environmentally adaptive multi-channel energy sensing path planning method for drone inspection of high-altitude photovoltaic power plants. Background Technology

[0002] Large-scale photovoltaic power plants typically consist of multiple long, strip-shaped rows of photovoltaic modules or several centralized subarrays. When using multi-rotor drones to perform autonomous inspection operations, in order to obtain infrared or visible light images that meet the requirements for image resolution and overlap, the drones must strictly adhere to preset altitudes, camera angles, and flight speeds to image along each photovoltaic row and make transfer flights between different photovoltaic rows or different subarrays.

[0003] Existing route planning methods for such inspection scenarios typically employ serpentine traversal of regular grids, nearest neighbor heuristic sorting, or Traveling Salesman Problem (TSP) models based on two-dimensional Euclidean distance. However, in photovoltaic power plants located in complex geographical environments such as high altitudes and mountains, the energy consumption of UAVs in the navigation environment exhibits highly anisotropic and non-uniform characteristics: First, the thin air at high altitudes leads to a significant decrease in atmospheric density, causing the hovering induced power and propulsion efficiency of multi-rotor UAVs to change non-linearly with absolute altitude; second, the undulating terrain of mountains causes UAVs to frequently generate large-amplitude climbs and descents when performing terrain-following routes, introducing a non-negligible increase in vertical power; third, the non-uniform wind field present in plateau stations causes flight segments with the same geometric displacement to produce drastically different horizontal relative air speeds (i.e., differences in drag due to tailwind, headwind, or crosswind) under different flight directions, resulting in completely asymmetrical directional energy costs for forward and reverse inspections, as well as bidirectional transfers between nodes.

[0004] In actual high-altitude mountain photovoltaic power stations, there are often widely distributed inverter booster equipment exclusion zones, lightning protection wire and tower safety buffer zones, undulating ridges, valleys, and local strong wind corridors. Under these complex constraints, there can be multiple feasible spatial channels with different geometric directions between any pair of inspection states. Among them, channels with shorter geometric distances may be forced to cross windward high-drag areas or terrain with strong climbs, while detour channels with slightly longer geometric distances may have lower overall flight energy consumption by utilizing local downwind or gentle terrain. If the system cannot explore and retain multiple feasible channels with different performance characteristics in parallel between state pairs, and only performs proportional energy consumption weighting based on distance on a single path, it will be unable to break the constraint of geometric distance on energy optimization, and it will be impossible to achieve multi-level collaborative energy perception optimization at the full mission scale between "global access order of photovoltaic rows", "intra-row inspection headings" and "inter-row three-dimensional transfer channels", making it difficult to adapt to the stringent airborne battery energy boundaries of complex high-altitude power stations.

[0005] In summary, existing technologies result in low accuracy and efficiency in environmentally adaptive multi-channel energy sensing path planning for UAV inspections of high-altitude photovoltaic power plants. Summary of the Invention

[0006] This invention provides an environment-adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants, which can improve the accuracy and efficiency of environment-adaptive multi-channel energy sensing path planning for UAV inspection of high-altitude photovoltaic power plants.

[0007] To achieve the above objectives, this invention provides an environment-adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants, comprising: The system acquires the digital elevation model, photovoltaic layout, hard exclusion zone, safety buffer zone, spatial wind vector field, and camera parameters of the photovoltaic power station to be inspected. Based on the tilt angle of the photovoltaic modules in the photovoltaic layout and the camera imaging geometric parameters in the camera parameters, the system constructs the spatial inspection trajectory and inspection speed limit for each photovoltaic row. For each photovoltaic array, a forward directional inspection state and a reverse directional inspection state with spatial geometric overlap but opposite flight directions are constructed. Based on the spatial position and flight direction of the flight segment, the absolute altitude of the midpoint of the flight segment is determined using a digital elevation model. The local air density at the absolute altitude is calculated based on a standard atmospheric model. The horizontal relative air velocity and vertical velocity components of the flight segment are determined by combining the spatial wind vector field. The free space defined by the digital elevation model, hard exclusion zone and safety buffer is discretized into a directed grid graph. The height of each grid node is set based on the digital elevation model elevation of the corresponding location and the preset transfer ground clearance. Each feasible directed edge in the directed grid graph is assigned a three-dimensional spatial geometric distance weight. The flight segment energy weight is dynamically updated according to the absolute altitude, spatial wind vector field and standard atmospheric model. Infeasible directed edges in the directed grid graph are removed according to the preset UAV flight performance constraints. For the relationship between the UAV base and each directed inspection state, as well as between any two different photovoltaic arrays, graph search is performed in the same directed grid graph based on three-dimensional spatial geometric distance weight and segment energy weight, respectively. The shortest distance channel and the lowest energy consumption channel are obtained and retained, and the range and channel energy consumption of the two types of channels are recorded. Four sets of directed connection matrices are generated to represent the range of the shortest distance channel, the energy consumption of flying along the shortest distance channel, the energy consumption of the lowest energy consumption channel, and the range of the lowest energy consumption channel. Using the access order of the photovoltaic rows in the outer traversal as the permutation variable, a two-state dynamic programming method is called to recursively decide the inspection direction of each photovoltaic row for the candidate access order in the permutation variable. The lowest energy consumption channel between adjacent directed inspection states is extracted from the four sets of directed connection matrices. Under the conditions of satisfying the preset battery available energy constraints, hard exclusion zone obstacle constraints and UAV flight performance constraints, the closed inspection route containing the access order of photovoltaic rows, the inspection direction of each photovoltaic row and the lowest energy consumption channel between photovoltaic rows is output.

[0008] To address the aforementioned problems, the present invention also provides an environment-adaptive multi-channel energy sensing path planning device for UAV inspection of high-altitude photovoltaic power plants, the device comprising: The environmental perception and inspection status modeling module is used to acquire the digital elevation model, photovoltaic layout, hard exclusion zone, safety buffer zone, spatial wind vector field, and camera parameters of the photovoltaic power station to be inspected. Based on the tilt angle of the photovoltaic modules in the photovoltaic layout and the camera imaging geometric parameters in the camera parameters, it constructs the spatial inspection trajectory and inspection speed limit corresponding to each photovoltaic row. For each photovoltaic row in the photovoltaic layout, it constructs forward directional inspection status and reverse directional inspection status with spatial geometric overlap but opposite flight directions. Based on the spatial position and flight direction of the flight segment as indexes, it uses the digital elevation model to determine the absolute altitude of the midpoint of the flight segment, calculates the local air density at the absolute altitude based on the standard atmospheric model, and determines the horizontal relative air velocity and vertical velocity components of the flight segment by combining the spatial wind vector field. The directed graph construction and multi-channel connection matrix solution module is used to discretize the free space defined by the digital elevation model, hard exclusion zone, and safety buffer into a directed grid graph. The height of each grid node is set based on the digital elevation model elevation of the corresponding location and the preset transfer ground clearance. Each feasible directed edge in the directed grid graph is assigned a three-dimensional spatial geometric distance weight, and the segment energy weight is dynamically updated according to the absolute altitude, spatial wind vector field, and standard atmospheric model. Infeasible directed edges in the directed grid graph are eliminated according to the preset UAV flight performance constraints. For the UAV base and each directed inspection state, as well as between any two different photovoltaic row units, a graph search is performed in the same directed grid graph based on the three-dimensional spatial geometric distance weight and the segment energy weight, respectively. The shortest distance channel and the lowest energy consumption channel are obtained and retained, respectively. The flight distance and channel energy consumption of the two types of channels are recorded, and four sets of directed connection matrices are generated to represent the shortest distance channel flight distance, the energy consumption along the shortest distance channel, the energy consumption of the lowest energy consumption channel, and the flight distance of the lowest energy consumption channel. The global route joint planning and output module uses the photovoltaic row access order of the outer traversal as the arrangement variable. For the candidate access order in the arrangement variable, it calls two-state dynamic programming to recursively decide the inspection direction of each photovoltaic row. It also extracts the lowest energy consumption channel between adjacent directed inspection states from four sets of directed connection matrices. Under the conditions of satisfying the preset battery available energy constraints, hard exclusion zone obstacle constraints and UAV flight performance constraints, it outputs a closed inspection route that includes the photovoltaic row access order, the inspection direction of each photovoltaic row, and the lowest energy consumption channel between photovoltaic rows.

[0009] To address the above problems, the present invention also provides an electronic device, the electronic device comprising: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to execute the above-described environmental adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants.

[0010] To address the aforementioned issues, the present invention also provides a computer-readable storage medium storing at least one computer program, which is executed by a processor in an electronic device to implement the aforementioned environmentally adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants.

[0011] This invention acquires the digital elevation model, photovoltaic (PV) array, hard exclusion zone, safety buffer zone, spatial wind vector field, and camera parameters of the photovoltaic power station to be inspected. Based on the tilt angle of the PV modules in the PV array and the camera imaging geometric parameters in the camera parameters, it constructs the spatial inspection trajectory and upper limit of the inspection speed for each PV array. This provides comprehensive geometric trajectory input and high-quality geometric configuration data. Furthermore, for each PV array, it constructs forward and reverse directional inspection states with spatial geometric overlap but opposite flight directions. Using the flight segment's spatial location and flight direction as indices, it determines the absolute altitude of the midpoint of the flight segment using the digital elevation model, calculates the local air density at that absolute altitude based on a standard atmospheric model, and combines this with... The spatial wind vector field determines the horizontal relative air velocity and vertical velocity components of the flight segment, which can improve the accuracy of energy consumption calculations caused by environmental anisotropy. Furthermore, the free space defined by the digital elevation model, hard exclusion zones, and safety buffer zones is discretized into a directed grid map. The height of each grid node is set based on the corresponding location's digital elevation model elevation and a preset transition ground clearance. Each feasible directed edge in the directed grid map is assigned a three-dimensional spatial geometric distance weight, and the flight segment energy weight is dynamically updated based on absolute altitude, spatial wind vector field, and a standard atmospheric model. Infeasible directed edges in the directed grid map are eliminated based on preset UAV flight performance constraints, preventing safety hazards such as flight paths crossing obstacles or inducing aerodynamic stall. In addition, regarding... Between the UAV base and each directed inspection state, and between any two directed inspection states of different photovoltaic arrays, graph searches are performed in the same directed grid graph based on three-dimensional spatial geometric distance weights and segment energy weights, respectively. The shortest distance channel and the lowest energy consumption channel are obtained and retained. The range and energy consumption of each type of channel are recorded, and four sets of directed connection matrices are generated representing the range of the shortest distance channel, the energy consumption along the shortest distance channel, the energy consumption of the lowest energy consumption channel, and the range of the lowest energy consumption channel. This allows for the parallel retention of both shortest distance and lowest energy consumption channels between the UAV base and each inspection state, and between directed inspection states of different photovoltaic arrays, generating four sets of directed connection matrices containing range and energy consumption information, thereby supporting... Subsequently, energy-priority and range-measurable connection decisions are made under asymmetric directed costs. Finally, the access order of the photovoltaic rows traversed from the outer layer is used as the permutation variable. For the candidate access order in the permutation variable, two-state dynamic programming is used to recursively decide the inspection direction of each photovoltaic row. The lowest energy consumption channel between adjacent directed inspection states is extracted from the four sets of directed connection matrices. Under the conditions of satisfying the preset battery available energy constraints, hard exclusion zone obstacle constraints, and UAV flight performance constraints, a closed inspection route containing the photovoltaic row access order, the inspection direction of each photovoltaic row, and the lowest energy consumption channel between photovoltaic rows is output. This can improve the accuracy and efficiency of environmentally adaptive multi-channel energy sensing path planning for UAV inspection of high-altitude photovoltaic power stations. Attached Figure Description

[0012] Figure 1 This is a flowchart illustrating an environmentally adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants, provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of the segment energy weight acquisition process of an environmental adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants, provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of the shortest path distance for an environment-adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants, provided in an embodiment of the present invention. Figure 4 This is a schematic diagram of the lowest energy consumption channel of an environmental adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants provided in an embodiment of the present invention. Figure 5 This is a functional block diagram of an environmentally adaptive multi-channel energy sensing path planning device for unmanned aerial vehicle (UAV) inspection of high-altitude photovoltaic power plants, provided in an embodiment of the present invention. Figure 6 This is a schematic diagram of the structure of an electronic device for implementing an environmentally adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants, as provided in an embodiment of the present invention.

[0013] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0014] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0015] This application provides an environment-adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants. The execution entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method can be executed by software or hardware installed on a terminal device or a server device, where the software may be a blockchain platform. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0016] Reference Figure 1 The diagram shown is a flowchart illustrating an environmentally adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants, provided by an embodiment of the present invention. In this embodiment, the environmentally adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants includes: It is understood that the embodiments of the present invention refer to obtaining the absolute altitude, local wind vector and altitude difference of each flight segment in the path planning stage, using the spatial location and flight direction of the flight segment as indexes, and calculating the local air density, horizontal relative air speed, vertical speed and energy weight of the flight segment accordingly; it does not specifically refer to real-time replanning during online machine learning or UAV flight.

[0017] S1. Obtain the digital elevation model, photovoltaic layout, hard exclusion zone, safety buffer zone, spatial wind vector field and camera parameters of the photovoltaic power station to be inspected, and construct the spatial inspection trajectory and inspection speed limit corresponding to each photovoltaic row based on the photovoltaic module tilt angle in the photovoltaic layout and the camera imaging geometric parameters in the camera parameters.

[0018] Understandably, a photovoltaic power station to be inspected refers to a physical site area where autonomous drone navigation monitoring, defect identification, or condition assessment is required. This photovoltaic power station is typically located in a high-altitude area, contains multiple arrayed photovoltaic module structures, and has operational environment characteristics such as undulating terrain, complex local micro-meteorology, and specific ground power facilities.

[0019] Understandably, photovoltaic (PV) layout refers to the geometric topology and arrangement information of all PV modules within a PV power station to be inspected in three-dimensional geographic space. It is usually represented in the form of discrete geometric entities, specifically covering the absolute spatial coordinates of each PV row, the position of the row centerline, the azimuth angle of the direction, the physical geometric length, the spacing between array rows, and the partitioning and subordinate relationships of each subarray.

[0020] Understandably, a hard exclusion zone refers to a three-dimensional geometric no-fly zone in space where entities are absolutely prohibited from crossing, flying over, or intruding during drone inspections and relocation flights. This includes, but is not limited to, tall ground-based energized facilities such as prefabricated substations, integrated inverter and booster platforms, power line towers, power transmission cable corridors, and meteorological observation towers, as well as collision avoidance safety buffer zones with specified spatial geometric tolerance distances.

[0021] Understandably, the spatial wind vector field refers to the vector dataset showing the distribution of local atmospheric flow patterns with three-dimensional spatial location within the three-dimensional airspace where the photovoltaic power station to be inspected is located. Its spatial wind vector field must at least contain two-dimensional or three-dimensional wind direction angles and wind speed distributions characterizing horizontal airflow. For example, the spatial wind vector field can be a uniform base wind field or a non-uniform wind field derived through computational fluid dynamics simulation / measurement inversion, such as containing local micrometeorological characteristics like specific gateway wind corridors, valley wind channels, or ridge acceleration zones.

[0022] Understandably, camera parameters refer to the optical and imaging hardware specifications of the inspection imaging sensor mounted on the drone gimbal, including but not limited to the physical focal length of the camera's optical lens, the physical size of the photosensitive element along the flight path, the image sensor resolution, the exposure time, and the shortest shooting interval for a single image determined by the shutter's mechanical / electrical performance. The inspection imaging sensor can be a visible light sensor or an infrared thermal imaging sensor.

[0023] Understandably, the tilt angle of a photovoltaic module refers to the spatial dihedral angle between the light-receiving surface of the photovoltaic panel (module) and the horizontal reference plane of the earth. This tilt angle directly determines the spatial orientation of the normal vector of the photovoltaic surface and is the fundamental geometric parameter for calculating the spatial relative displacement of the inspection route when the UAV maintains a straight or constant tilt angle for observation.

[0024] It is understandable that camera imaging geometry parameters refer to the pre-set optical observation spatial geometry relationship and overlap rules to meet the requirements of image resolution (Ground Sample Distance, GSD) and spatial coverage integrity of the inspection images. These include at least the normal spatial straight-line distance that the UAV camera lens needs to maintain relative to the light-collecting surface of the photovoltaic module, and the minimum forward overlap rate threshold of two consecutive exposure images on the flight path.

[0025] Understandably, a photovoltaic (PV) row refers to a long, single physical PV support structure unit consisting of multiple PV modules connected in series and parallel and arranged linearly along a specific direction. In path planning, a single PV row serves as the smallest independent discrete linear geometric target covered by the inspection task, and in the planning logic, it corresponds to two possible unidirectional traversal inspections along the opposite extension direction of the row.

[0026] Understandably, the spatial inspection trajectory refers to the continuous three-dimensional flight path that the center of the drone or the optical center of the camera lens must strictly follow in three-dimensional space during the process of performing effective ground imaging operations on a designated photovoltaic array. This trajectory is usually generated by spatially translating the spatial centerline of the photovoltaic array along the normal spatial distance of the light-collecting surface of the module by a preset normal distance, maintaining geometric parallelism with the surface of the photovoltaic array module to ensure that the shooting line of sight is orthogonal and maintains a constant object distance.

[0027] Understandably, the inspection speed limit refers to the maximum horizontal ground flight speed allowed for the UAV to move relative to the surface of the photovoltaic array per unit time during in-array inspection imaging flight, in order to ensure that the forward overlap rate of the captured images in the flight direction is not lower than a preset safe overlap threshold. This inspection speed limit is calculated analytically from the camera's shortest shooting interval, the sensor's directional size, focal length, and normal shooting distance, and is used to limit the maximum speed boundary of actual flight within the array to avoid missed shots or insufficient overlap.

[0028] It is understood that the embodiments of the present invention also need to obtain the battery parameters that provide airborne flight propulsion and power supply for airborne equipment of the UAV. The battery parameters are the electrical and energy boundary attributes of the power battery pack, including but not limited to the nominal terminal voltage, rated total energy storage, discharge energy conversion efficiency of the power battery, and the low-power safety reserve threshold set in advance to ensure that the UAV can return and land safely, that is, the maximum available battery energy for the mission determined after deducting the reserve.

[0029] It is understood that the embodiments of the present invention also need to obtain the flight parameters of the UAV. The flight parameters of the UAV refer to the inherent dynamics, kinematics and physical structure properties of the multi-rotor UAV platform performing the inspection operation, including but not limited to the total takeoff mass of the UAV, the total effective rotation area of ​​the rotor disk, the aerodynamic drag coefficient in each direction, the energy conversion efficiency of the motor and ESC, the nominal cruise speed during the transition, the maximum permissible climb speed, the maximum permissible descent speed, the maximum wind resistance limit, and the fixed static power consumption of the flight control / communication avionics equipment.

[0030] Specifically, based on the tilt angle of the photovoltaic modules in the photovoltaic array and the camera imaging geometry parameters in the camera parameters, the spatial inspection trajectory and upper limit of the inspection speed for each photovoltaic array are constructed, including: The center line of the photovoltaic array is translated along the normal direction of the photovoltaic module surface by a preset camera normal distance to generate a spatial inspection trajectory parallel to the photovoltaic module surface; The maximum inspection speed is calculated based on the camera normal distance, sensor physical dimensions along the flight path, camera focal length, camera minimum shooting interval, and forward overlap rate. The actual inspection speed is then limited to no more than the maximum inspection speed.

[0031] Understandably, the preset camera normal distance refers to the spatial orthogonal straight-line geometric distance between the optical center (or effective imaging plane) of the UAV's onboard camera and the flat surface of the photovoltaic module's light-collecting surface, which is pre-calibrated along the spatial normal vector perpendicular to the light-collecting surface and must be strictly maintained throughout the entire process during the in-row inspection imaging operation.

[0032] Understandably, in this embodiment of the invention, the camera normal distance is preset as the object distance control benchmark, which matches the camera focal length and image sensor pixel size, directly determining the ground sampling distance (GSD) and the effective field of view (FOV) of the image. By offsetting this distance in the normal direction along the center line of the photovoltaic array in three-dimensional space, the three-dimensional translational spatial route that the body center must follow during the UAV inspection flight can be calculated, so as to ensure that the camera can maintain a constant normal line of sight orthogonal observation and a constant imaging spatial resolution on the photovoltaic surface during flight.

[0033] Understandably, the forward overlap rate refers to the percentage of the geometric length of the ground photovoltaic module space area covered by the latter photo along the flight path (the direction of the flight path extension) that overlaps with the space area covered by the former photo in two consecutive inspection photos taken by the UAV along the space inspection trajectory, relative to the total ground coverage length of a single photo.

[0034] It is understood that the forward overlap rate in the embodiments of the present invention is the lower margin threshold that must be achieved to ensure the localization of photovoltaic defect images, multi-viewpoint geometric registration, and the integrity of orthophoto stitching. When the forward overlap rate requirement is increased, the spatial baseline distance between photos is shortened accordingly. Therefore, when limited by the shortest exposure time interval of the camera hardware, this indicator constitutes the rigid geometric and temporal upper limit boundary for the reverse analysis derivation of the maximum allowable inspection flight speed of the UAV within the row.

[0035] S2. For each photovoltaic array, construct forward and reverse directional inspection states with spatial geometric overlap but opposite flight directions. Based on the spatial position and flight direction of the flight segment, use the digital elevation model to determine the absolute altitude of the midpoint of the flight segment, calculate the local air density at the absolute altitude based on the standard atmospheric model, and combine the spatial wind vector field to determine the horizontal relative air velocity and vertical velocity components of the flight segment.

[0036] It is understood that, after calculating the local air density at absolute altitude based on the standard atmospheric model and determining the horizontal relative air velocity and vertical velocity components of the flight segment by combining the spatial wind vector field, the embodiments of the present invention also include performing directional energy integration along each spatial inspection trajectory to obtain the inspection energy consumption corresponding to each forward directional inspection state and the reverse directional inspection state.

[0037] Understandably, the spatial location of a flight segment refers to the absolute geometric location distribution of the various micro-element flight segment units formed after the spatial inspection trajectory is discretized in three-dimensional geographic space. In the process of numerical calculation and parameter query, it is usually quantitatively represented by the three-dimensional Cartesian spatial coordinates or geodetic coordinates (including plane coordinates and elevation) of the geometric center point of the micro-element flight segment, which serves as a global index benchmark for extracting local terrain elevations from the Digital Elevation Model (DEM) and sampling local micro-meteorological data from the spatial wind vector field.

[0038] Understandably, flight direction refers to the spatial motion pointing vector of a UAV when flying along a specific flight path element, i.e., the heading angle and pitch angle. In the construction of directional inspection states, although the forward and reverse inspection states corresponding to the same physical photovoltaic array have overlapping spatial three-dimensional geometric line segments, the velocity vector directions at corresponding points on these line segments are strictly opposite. Flight direction directly determines the spatial dihedral angle and included angle projection between the UAV's heading and the local spatial wind vector. It is the core vector feature that distinguishes the differences in aerodynamic drag when flying with the wind, against the wind, or across the wind, and introduces the cost calculation when asymmetric directional flight is introduced.

[0039] Understandably, the standard atmospheric model refers to a mathematical, empirical, and thermodynamic physical model used to characterize the relationship between the physical properties of the Earth's atmosphere and absolute altitude. In this embodiment of the invention, the standard atmospheric model typically takes sea-level standard reference parameters (including sea-level standard temperature and sea-level standard pressure) and the standard temperature lapse rate in the troposphere as inputs. It sequentially derives and calculates the local ambient temperature and local static pressure from the absolute altitude, and then establishes the equation of state by combining gravitational acceleration and the dry air gas constant, outputting the atmospheric thermodynamic state parameters at that specific absolute altitude.

[0040] Understandably, local air density refers to the physical mass of a unit volume of atmosphere at the actual absolute altitude, temperature, and air pressure at the midpoint of a specified space segment. Due to the high absolute altitude and low air pressure, the local air density of high-altitude photovoltaic power stations is significantly lower than that at standard sea level. This directly affects the aerodynamic thrust generated by the rotor rotation of multi-rotor UAVs, the induced downwash velocity, and the rotor profile drag power. It is a key state variable determining the nonlinear increase in propulsion energy consumption of UAVs in high-altitude environments.

[0041] Understandably, horizontal relative airspeed refers to the spatial vector difference between the UAV's velocity vector in the horizontal plane (i.e., the horizontal ground speed vector relative to the ground) and the horizontal wind vector at the midpoint of the UAV's flight segment (i.e., the geometrically combined vector of relative wind direction and relative wind speed, i.e., the projection of the airspeed vector onto the horizontal plane). This airspeed vector objectively reflects the actual aerodynamic angle of attack and relative airflow scouring speed experienced by the UAV's airframe and rotor, directly determining the amount of horizontal flight propulsion power required to overcome parasitic drag and induced drag.

[0042] Understandably, the vertical velocity component refers to the magnitude and sign of the velocity of the UAV moving along the vertical axis of the three-dimensional coordinate system of the earth. When performing three-dimensional terrain following inspection and relocation along undulating terrain or component slopes, the vertical velocity component reflects the rate of motion displacement of the UAV due to altitude climb (requiring additional work to overcome gravitational potential energy) or altitude descent (potential energy release but requiring prevention of the rotor from entering a vortex ring state). It serves as an independent power calculation channel to calculate the vertical motion power increment during climb and descent.

[0043] Understandably, directional energy integration refers to the numerical calculation process of multiplying the instantaneous total flight power of each discrete segment along a specified flight path by the corresponding flight time, and then numerically accumulating (summing or discretely integrating) the instantaneous energy consumption of all segments along the topological direction of the trajectory to convert it into standard energy units. Since the instantaneous total power is determined by both the horizontal relative airspeed and the vertical component, which are strongly correlated with direction, the total energy cost obtained by integrating from the tip to the tip (forward) and from the tip to the tip (reverse) usually exhibits direction dependence and asymmetry when wind field or climb / descent differences exist.

[0044] Understandably, inspection energy consumption refers to the total electrical energy consumed by the UAV's power system and onboard payload system during the unidirectional complete imaging flight of a single photovoltaic array. This inspection energy consumption is obtained by directional energy integration along the inspection trajectory, specifically including the aerodynamic propulsion energy consumption of the onboard multi-rotor used to provide vertical lift and horizontal propulsion, the power increment for terrain and attitude maintenance, and the sum of the fixed static energy consumption generated by the flight control and navigation system, the onboard data communication link module, and the inspection camera hardware performing continuous ground imaging during the inspection time period.

[0045] For example, the standard atmospheric temperature at an absolute altitude h is calculated using the following formula through a standard atmospheric model: ; Where T(h) represents the standard atmospheric temperature at an absolute altitude h, T0 represents the standard sea-level temperature, and L represents the temperature lapse rate. h is the altitude of the midpoint of the three-dimensional short segment relative to a unified altitude benchmark, in meters (m), with temperature in Kelvin (K), and L in K / m.

[0046] For example, the standard atmospheric pressure at an absolute altitude h is calculated using the following formula through a standard atmospheric model: ;in, Absolute altitude Standard atmospheric pressure at that location Standard atmospheric pressure at sea level. altitude The absolute temperature at that location The standard temperature at sea level It is the acceleration due to gravity. The gas constant of air. For temperature lapse rate, and The unit is Pa. The unit is m / s 2 , The unit is J / (kg·K).

[0047] For example, the local air density at absolute altitude is calculated using the following formula: ;in, Absolute altitude The local air density.

[0048] S3. Discretize the free space defined by the digital elevation model, hard exclusion zone and safety buffer into a directed grid graph. The height of each grid node is set based on the digital elevation model elevation of the corresponding location and the preset transfer ground clearance. Assign a three-dimensional spatial geometric distance weight to each feasible directed edge in the directed grid graph. Dynamically update the segment energy weight according to the absolute altitude, spatial wind vector field and standard atmospheric model. Eliminate infeasible directed edges in the directed grid graph according to the preset UAV flight performance constraints.

[0049] Understandably, a safety buffer zone refers to a three-dimensional protective isolation zone formed by extending outwards in three-dimensional space with a predetermined safety redundancy margin (such as horizontal extension and vertical elevation by a specific safety distance) based on the geometric outline of the hard exclusion zone defined by the digital elevation model and site facilities. This safety buffer zone, together with the hard exclusion zone, constitutes an insurmountable prohibited entity, used to absorb GPS positioning drift, flight control trajectory tracking errors, and aerodynamic disturbances caused by rotor downwash under strong winds at high altitudes, ensuring that the drone's flight path does not physically scrape or collide with ground-based inverters, transformer substations, iron towers, or photovoltaic supports.

[0050] Understandably, a directed grid graph refers to a spatial topology network model constructed by discretizing a three-dimensional free-flying space according to a preset resolution (such as a specific grid step size) in a regular or irregular manner. This graph consists of discrete three-dimensional spatial location nodes and directed edges (such as eight-neighbor topology edges) that connect adjacent nodes with strict spatial orientation. Between any two nodes in the graph, there are two independent directed edges with opposite directions, used to explicitly bear the asymmetric transfer costs caused by differences in wind direction, ascent, and descent.

[0051] Understandably, the preset transfer altitude refers to the minimum vertical clearance that the UAV's underside must maintain throughout its flight path relative to the actual ground surface (or the surface of ground vegetation or low obstacles) when performing spatial transfer flights between photovoltaic arrays or between photovoltaic arrays and power station bases. The three-dimensional absolute altitude of each grid node is generated by directly superimposing this preset transfer altitude onto the elevation values ​​of the digital elevation model (DEM) at the corresponding planar location, thus giving the grid network an overall three-dimensional terrain-following spatial surface shape that ensures safe clearance.

[0052] Understandably, the 3D spatial geometric distance weight refers to the actual geometric line segment length in 3D Cartesian Euclidean space corresponding to the directed connection edge from the starting grid node to the ending grid node in a directed grid graph. This 3D spatial geometric distance weight depends only on the geometric difference of the 3D spatial coordinates of the two nodes, and is used to represent the pure spatial Euclidean geometric span cost between the two points in graph search. It is independent of the UAV's airframe dynamics, ambient wind speed, and flight direction (i.e., the geometric distance weights of the forward and reverse edges are strictly equal).

[0053] Understandably, segment energy weight refers to the scalar energy consumption value obtained by productting the instantaneous total flight power corresponding to the directed connecting edge (a combination of induced power base determined by local air density, horizontal relative airspeed power, vertical motion power increment, turning additional power, and auxiliary airborne equipment power consumption) with the motion time required for the UAV to fly along that connecting edge. Since relative airspeed and vertical motion have directional polarity, the segment energy weights assigned to two directed edges with opposite directions between the same two nodes are usually not equal.

[0054] Understandably, UAV flight performance constraints refer to rigid kinematic and dynamic boundary conditions jointly determined by the limits of the multi-rotor UAV's power mechanism, inherent aerodynamic characteristics, and flight safety boundaries. These include, but are not limited to, the maximum permissible vertical climb speed, the maximum permissible vertical descent speed (to avoid the rotor entering a dangerous vortex ring or severe flutter state), the maximum permissible horizontal flight speed, the maximum permissible upper limit of the environmental wind resistance speed, and the lower limit of the effective applicability range of the empirical model for multi-rotor descent propulsion energy. Any candidate directed edge exceeding the above boundary conditions is deemed physically infeasible and rigidly eliminated during the graph construction phase.

[0055] It is understood that the embodiments of the present invention dynamically update the segment energy weight based on absolute altitude, spatial wind vector field and standard atmospheric model, and the segment energy weight can be dynamically updated by calculating absolute altitude, local wind vector, vertical motion and turning attitude.

[0056] Understandably, absolute altitude refers to the vertical absolute geometric height of a three-dimensional micro-element segment represented by a directed grid edge, at the geometric midpoint of the segment, relative to a unified geoid (such as the Yellow Sea Height Datum or the WGS-84 ellipsoid). This parameter differs from the ground clearance relative to the ground and is directly used as a global input to the standard atmospheric physics model to calculate the atmospheric static pressure, local temperature, and local air density at that altitude.

[0057] Understandably, a local wind vector refers to a spatial vector obtained by sampling the three-dimensional absolute coordinates of the midpoint of a micro-element flight segment from the spatial wind vector field, representing the instantaneous intensity and direction of atmospheric flow within a specific local airspace. In the horizontal coordinate system, it is decomposed into vector components containing the magnitude of the horizontal wind speed and the horizontal wind direction angle, directly reflecting the local micro-meteorological airflow characteristics that cause aerodynamic scouring of the UAV's flight within that small local area.

[0058] Understandably, vertical motion refers to the ascent or descent gliding maneuvers that a UAV must perform along the vertical zenith axis during flight along a directed connecting edge, due to the absolute elevation difference between the starting and ending nodes. This vertical motion directly determines the rate of change of vertical displacement (vertical velocity component) along the flight path, which corresponds in energy consumption calculations to the increase in climbing power required to overcome gravitational potential energy, or the adjustment of descent power to release potential energy under restricted and controlled conditions.

[0059] Understandably, turning attitude refers to the spatial tilt, roll, and yaw attitude changes that occur when a UAV changes its motion vector direction from the previous segment to enter the current segment, in order to provide the centripetal acceleration required for turning. The change in turning attitude causes the total thrust vector generated by the multirotors to deviate from the zenith direction. In order to maintain the original lift and flight path altitude during tilting turns, the rotor system must output an additional thrust multiplier, thus inducing a non-negligible additional thrust power for turning in this segment.

[0060] Specifically, the energy weighting of flight segments is dynamically updated based on absolute altitude, spatial wind vector field, and standard atmospheric model, including: The spatial trajectory is discretized into flight segment units. The absolute altitude of the midpoint of the flight segment unit is input into the standard atmospheric model. The local temperature is calculated from the standard sea level temperature and the temperature lapse rate. The local air pressure is calculated from the standard sea level pressure, local temperature, gravitational acceleration, and air gas constant. Then the local air density is calculated. The ideal induction velocity for hovering is calculated from the local air density, and the profile power reference is scaled. The horizontal wind vector at the midpoint of the flight segment is sampled from the space wind vector field. The horizontal relative air velocity is determined by the difference between the UAV's horizontal ground speed vector and the horizontal wind vector. The horizontal flight power component is then calculated. Calculate the vertical motion power increment based on the vertical velocity components of the flight segment; For turning segments with a preset finite turning radius, the thrust multiple is determined based on the horizontal speed and the turning radius, and the additional turning power is calculated. For straight grid edges in the directed grid diagram, the additional turning power is set to zero. The instantaneous total power is obtained by adding the horizontal flight power component, vertical motion power increment, turning additional power, fixed power consumption of avionics and communication, and camera power consumption. The directional energy weight that varies with the spatial position and flight direction of the flight segment is generated based on the instantaneous total power and the flight time integral of the segment.

[0061] For example, the horizontal wind vector at the midpoint of the sampling segment cell from the spatial wind vector field is calculated, and the horizontal relative air velocity is determined by the difference between the UAV's horizontal ground velocity vector and the horizontal wind vector, using the following formula: ; in, The horizontal relative air velocity, It is the horizontal ground velocity vector. The horizontal wind vector is the midpoint of the flight segment unit.

[0062] For example, the ideal hovering induction velocity is calculated from the local air density using the following formula: ; in, This indicates the ideal hovering induced speed. Indicates the total weight of the drone. This indicates the total rotor disk area.

[0063] For example, the ideal hovering induction velocity is calculated from the local air density, and the profile power reference is scaled using the following formula: ; in, This represents the profile power reference at the current weight and air density. Indicates the power reference of the reference state profile. For reference total weight, For reference air density.

[0064] For example, the vertical motion power increment is calculated based on the vertical velocity components of the flight segment using the following formula: ; in, Indicates the increase in vertical motion power. The axial induced velocity, For vertical speed of the flight segment, ;in, Indicates the axial induced velocity. Indicates the vertical speed of the flight segment. This indicates the ideal hovering induced speed.

[0065] For example, the horizontal flight power component is calculated using the following formula: ;in, Represents the horizontal flight power component. This represents the induced power correction factor. Represents the equivalent resistance area. Forward flight induced power factor, The magnitude of the relative air velocity. Let be the equivalent tip velocity in the current state, where ,in, This represents the equivalent tip velocity in the current state. Indicates the reference equivalent tip velocity, where, ;in, This represents the forward flight induced power factor. Indicates the magnitude of relative air velocity. This indicates the ideal hovering induced speed.

[0066] For example, the additional power during turning is calculated using the following formula: ; in, Indicates additional power during turning. This represents the induced power at the current horizontal relative air velocity. This indicates the thrust multiple determined by the centripetal acceleration during turning.

[0067] It is understood that, in existing implementations, the turning additional power of the straight grid edge is zero when the turning radius is infinitely large; turning additional power is only calculated for turning segments with finite turning radii.

[0068] For example, the instantaneous total power is calculated using the following formula: ;in, This indicates the instantaneous total power of the drone. This indicates the fixed power consumption of flight control and communication systems. This indicates the camera's power consumption.

[0069] S4. For the UAV base and each directed inspection state, as well as for any two different photovoltaic rows, perform graph search in the same directed grid graph based on three-dimensional spatial geometric distance weight and segment energy weight respectively, obtain and retain the shortest distance channel and the lowest energy consumption channel respectively, record the range and channel energy consumption of each type of channel, and generate four sets of directed connection matrices representing the range of the shortest distance channel, the energy consumption of flying along the shortest distance channel, the energy consumption of the lowest energy consumption channel, and the range of the lowest energy consumption channel.

[0070] Understandably, graph search refers to the numerical optimization process in a directed grid graph network model composed of grid nodes and feasible directed edges. Starting from a specific graph node (such as a UAV base node or an exit grid node of a certain directed inspection state), the search explores the topological network using graph theory optimal path traversal and relaxation algorithms according to preset optimal target weights (three-dimensional spatial geometric distance weights or flight segment energy weights) to find the directed node / edge connection sequence with the minimum cost in the corresponding weight dimension from the starting point to the specified target node (such as the entrance grid node of each candidate directed inspection state or the landing node returning to the base). The graph theory optimal path traversal and relaxation algorithms can adopt single-source shortest path algorithms, heuristic search algorithms, etc.

[0071] Understandably, range and channel energy consumption refer to the cumulative three-dimensional geometric flight distance and the total electrical energy consumed by the entire aircraft's propulsion and onboard systems when the UAV performs a three-dimensional spatial transfer flight in a specific spatial connection channel calculated by the directed grid diagram.

[0072] Understandably, a directed connection matrix is ​​a two-dimensional bidirectional numerical lookup table used in the high-level planning of inspection tasks to store the transition connection cost values ​​between the source state (including the base and the exit nodes of each directed inspection state) and the target state (including the entry nodes of each row of directed inspection states and the base).

[0073] Understandably, the two types of channels refer to two sets of spatial 3D topology transfer routes independently searched and retained in parallel based on different optimization objective criteria, for the same pair of directed inspection state nodes (or between base and state) in the same underlying feasible directed grid graph. Specifically, these include: 1. Shortest path: A connection path with the global minimum three-dimensional Euclidean geometric length obtained by searching using three-dimensional spatial geometric distance weight as a single optimization cost function; 2. Lowest Energy Consumption Path: The connection path with the lowest global energy consumption is obtained by searching using the segment energy weight (coupled with high altitude and low density, spatial wind field vector relative airspeed, terrain undulation climbing and descending power, and turning attitude) as a single optimization cost function. When the two types of paths detour through strong wind areas, valley terrain, or exclusion zones, they usually exhibit different geometric polygonal trajectories and directions in three-dimensional physical space (when the space overlaps in a specific windless, flat, and open environment, they are also recorded and retained independently with different costs), making the selection of the transfer path a high-level spatial decision variable independent of the inspection team's access order.

[0074] Understandably, after recording the range and energy consumption of each of the two types of channels and generating four sets of directed connection matrices representing the range of the shortest channel, the energy consumption of flying along the shortest channel, the energy consumption of the lowest energy consumption channel, and the range of the lowest energy consumption channel, the embodiments of the present invention further include associating the directed connection matrix of the shortest channel with the path recovery information of the shortest channel; and associating the directed connection matrix of the lowest energy consumption channel with the path recovery information of the lowest energy consumption channel.

[0075] Understandably, path recovery information refers to a set of precursor indication data synchronously stored with the matrix data, in order to seamlessly reconstruct the continuous geometric topology trajectories of the two types of channels in three-dimensional physical space while obtaining the globally optimal cost value during graph search, and in order to achieve this in subsequent task calculation and route generation stages. Path recovery information includes, but is not limited to, the parent node pointers of each intermediate grid node recorded during single-source shortest path relaxation, the precursor directed edge index sequence, the adsorbed connection micro-elements, or the discrete three-dimensional space waypoint coordinate sequence generated by the reverse backtracking calculation and expansion of the above precursor sequences. By calling this information, a single cost value value in the matrix can be seamlessly mapped and restored to a continuous three-dimensional flight path segment that can be directly executed by the UAV flight control system.

[0076] Specifically, the free space defined by the digital elevation model, hard exclusion zone, and safety buffer is discretized into a directed raster graph, including: The directed grid graph is set to an eight-neighbor topological connection, with both endpoints of the connection edge located in free space, and the two orthogonal adjacent nodes corresponding to the diagonal connection edge located in free space to prevent corner points from crossing. The flight time and velocity components in each direction are determined based on the three-dimensional geometric length of the connecting edges and the set transfer speed. Directed edges that violate the maximum climb speed limit, maximum descent speed limit, maximum horizontal flight speed limit, local maximum wind speed limit, or the applicable range limit of the descent propulsion model are eliminated.

[0077] For example, the flight time and velocity components in each direction are determined based on the three-dimensional geometric length of the connecting edge and the set transfer speed, and are calculated using the following formula: ;in, Represents a node To the node The three-dimensional geometric length, Represents a directed edge arrive Energy rights for flight segments and Represents the three-dimensional coordinates of the corresponding node. This represents the instantaneous total power of the directed edge. Indicates the transition speed.

[0078] Understandably, eight-neighbor topology connection refers to a directed network connection method that establishes candidate topology connections between each grid node in a two-dimensional projected grid plane and its eight adjacent grid nodes (including four orthogonal adjacent nodes to the east, west, south, and north, and four diagonal adjacent nodes to the northeast, southeast, northwest, and southwest). This topology construction mechanism introduces strict corner-cutting prevention geometric constraints: a diagonal connection edge is only allowed to be retained as a feasible connection edge in the graph if both endpoints (starting and ending nodes) of that diagonal connection edge are in free space, and both orthogonal lateral adjacent nodes spanned by that diagonal connection edge are also in free space. If either of the two orthogonal adjacent nodes falls into an obstacle zone or a safety buffer zone, the diagonal connection edge is forcibly blocked and removed. This rigidly prevents geometric safety hazards caused by UAVs cutting directly along the diagonal edges of hard exclusion zones or safety buffer zones at corners, thus eliminating the risk of collisions.

[0079] Understandably, free space refers to the set of three-dimensional, continuously connected physical airspace remaining after removing all hard exclusion zones (including the outlines of ground-based inverter rooms, prefabricated substations, transmission towers, central control rooms, and obstacles such as tall trees) and their outward-expanding safety buffer zones along the three-dimensional space, which allows UAVs to legally and safely hover and freely cruise. In the rasterized discrete representation, free space is represented as the set of grid nodes and feasible navigation lines whose geometric three-dimensional spatial coordinates do not fall within any obstacle entity or safety collision avoidance redundant envelope.

[0080] Understandably, the limitation of the applicable range of the descent propulsion model refers to the physical boundary of the working conditions in which the mathematical or momentum-blade element aerodynamic propulsion model is allowed to be effectively used for analytical calculation of the rotor propulsion energy consumption of multi-rotor UAVs with a vertical downward motion component (negative vertical velocity). When a multi-rotor UAV flies at an excessively high vertical descent rate, the high-speed downwash airflow below the rotor is prone to upward turbulence and entrainment into the rotor blade tips, inducing a dangerous vortex ring state. At this time, the conventional propulsion power approximation model will exhibit non-physical low energy consumption or even negative power distortion, resulting in complete distortion and failure. This limitation refers to the rigid filtering and removal of all directed edges in the directed graph construction stage that exceed the effective fitting and engineering prediction range of the empirical propulsion energy model or induce model distortion and rotor aerodynamic instability risk, as unreliable / unsafe invalid edges. This ensures the physical authenticity and numerical calculation robustness of the energy weights for the entire flight segment. The vertical descent rate is the ratio of the vertical descent speed to the ideal induced speed for rotor hovering. The descent ratio threshold exceeds the limit calibration range of the model, such as the upper limit of the descent ratio given in the embodiment of this invention, or a non-physical singular condition with a high descent rate that causes the momentum theory induced speed solution to fail in the power calculation.

[0081] Furthermore, graph search is performed on the same directed grid graph based on three-dimensional spatial geometric distance weights and segment energy weights, respectively, to obtain and retain the shortest distance channel and the lowest energy consumption channel. The range and energy consumption of each type of channel are recorded, and four sets of directed connectivity matrices are generated to represent the range of the shortest distance channel, the energy consumption along the shortest distance channel, the energy consumption of the lowest energy consumption channel, and the range of the lowest energy consumption channel. These include: For each exit node of a directed inspection state, a single-source shortest path search is performed with three-dimensional spatial geometric distance weight and segment energy weight respectively, to obtain the shortest distance channel and the lowest energy consumption channel to each target directed inspection state entry node respectively. The search results are stored in the corresponding directed connection matrices, and the two types of channels are associated with the predecessor node or the predecessor edge sequence, respectively. When the shortest path and the lowest energy consumption path between the same state pair coincide in spatial geometry, they share the same path sequence and record the corresponding travel cost and energy consumption cost respectively.

[0082] Understandably, a predecessor node refers to the parent node (or pointer) that, in the shortest path search based on graph theory (such as single-source shortest path search), is the immediate neighbor directly in front of the current relaxed grid node on the optimal path starting from the source node (i.e., the grid node associated with a directed inspection state exit) and directly undergoes a single-step directed topology transition to it. After the graph search converges, by recursively backtracking from the target state entry grid node along each level of predecessor node pointers to the source state exit node, the discrete three-dimensional grid node spatial coordinate sequence constituting the optimal connection channel can be deterministically extracted.

[0083] Understandably, the precursor edge sequence refers to the set (or edge index identifier sequence) of ordered directed edges (or a set of edges) with strict geometric orientation extracted sequentially from adjacent nodes during the process of searching for the shortest path from a single source and backtracking the path. Since each feasible directed edge is directly bound to calculated parameters such as three-dimensional spatial geometric vectors, local air density, wind direction angle, relative air speed, and segment energy weights in the underlying grid network, the precursor edge sequence constitutes a complete spatial connection chain that can directly and losslessly reconstruct the dynamic parameters and three-dimensional topological morphology of the micro-element segment.

[0084] Understandably, range cost refers to the algebraic scalar sum (usually in meters) of the three-dimensional geometrical physical lengths (whether the channel is the shortest distance channel generated by geometric distance or the lowest energy consumption channel generated by the energy model) of all infinitesimal segments contained in the three-dimensional Cartesian Euclidean space during a UAV's flight transition along a specific connecting channel (whether the channel is the shortest distance channel generated by geometric distance or the lowest energy consumption channel generated by the energy model). This value is used to objectively quantify the total pure three-dimensional spatial displacement distance paid by the UAV in completing the spatial transition between two states in the spatial geometric dimension, and is a benchmark cost indicator for evaluating the compactness of the flight path and the geometrical path length.

[0085] Understandably, energy consumption cost refers to the total electrical energy (usually measured in joules or watt-hours) consumed by the UAV's multi-rotor propulsion system (used to overcome low-pressure induced power drag, horizontal relative wind resistance, terrain-following climb work, and additional power for turning and tilting) and onboard electrical load systems (flight control, communication, and auxiliary equipment) during the entire flight transfer process along a specific connecting channel. In system implementation, this cost is obtained by multiplying the instantaneous total power on each infinitesimal flight segment contained in the channel by the flight time and then summing the results by discretely integrating segment by segment along the topological direction. This index represents both the global minimum energy transfer achieved when flying along the "lowest energy consumption channel" and is used to accurately record the actual energy consumption when flying along the "shortest distance channel," thus forming a cross-dimensional quantitative evaluation of range and energy in four sets of directed connection matrices.

[0086] S5. Using the access order of the photovoltaic rows in the outer traversal as the permutation variable, the two-state dynamic programming is called to recursively decide the inspection direction of each photovoltaic row for the candidate access order in the permutation variable. The lowest energy consumption channel between adjacent directed inspection states is extracted from the four sets of directed connection matrices. Under the conditions of satisfying the preset battery available energy constraints, hard exclusion zone obstacle constraints and UAV flight performance constraints, the closed inspection route containing the access order of photovoltaic rows, the inspection direction of each photovoltaic row and the lowest energy consumption channel between photovoltaic rows is output.

[0087] Understandably, outer-layer traversal refers to the macroscopic search process in which discrete permutations of photovoltaic (PV) rows are used as the main control variables of the outer loop in the entire hierarchical path planning architecture. Through a combination optimization traversal mechanism, different candidate permutation sequences are systematically generated, evaluated, and iteratively improved in the permutation search space composed of all physical PV rows. When the outer layer provides a global PV row topology sequence in each iteration, the inner-layer deterministic optimization algorithm (two-state dynamic programming and channel extraction) is synchronously triggered and called to calculate the low-energy cost of the algorithm.

[0088] Understandably, the photovoltaic (PV) row access order refers to the sequential order in which drones fly over and cover all designated physical PV rows within the PV power station to be inspected. In the mathematical programming model, this is specifically represented as a permutation of all ordered discrete sets (Permutation) consisting of the physical numbers (or unique identifiers) of the PV rows. This variable only macroscopically defines which PV row the drone flies from to the next PV row in sequence, and has not yet been bound to the specific entry and exit directions and entry / exit status within each row.

[0089] Understandably, the battery available energy constraint refers to the total electrical energy consumed by the onboard power system, control system, and payload during the entire mission of a complete inspection route by the UAV. This energy must strictly not exceed the maximum allowable energy consumption limit after deducting the preset safety reserve capacity from the rated total capacity of the UAV's onboard power battery pack. This constraint constitutes a rigid inequality criterion for screening and determining whether candidate inspection routes possess engineering and physical feasibility in terms of dynamics and energy safety.

[0090] Understandably, the hard exclusion zone constraint means that during the entire inspection operation (including in-squad inspection flights, inter-squad airspace transfers, and the entire flight path from takeoff to arrival and return to base), the three-dimensional spatial geometric coordinates of the UAV's center, rotor sweep envelope, and outer contour at any given moment are strictly restricted to the impassable entity areas defined by the geometric outer contours of each hard exclusion zone and its three-dimensional safety buffer zone. This constraint requires that the flight path never interfere with, cross, or intrude into any three-dimensional spatial entities to ensure absolute spatial collision avoidance safety for ground facilities.

[0091] Understandably, flight performance constraints refer to the requirement that the instantaneous kinematics and dynamics of the UAV on any micro-segment of the entire mission route must strictly remain within the pre-set physical limits and safe flight envelope boundaries of the UAV platform. This includes, but is not limited to: the vertical ascent rate of the route must not exceed the set maximum permissible climb rate; the vertical descent rate of the route must not exceed the set maximum permissible descent rate (to avoid rotor downwash being drawn into the vortex ring, leading to steady-state loss of control); the horizontal cruise speed must not exceed the maximum permissible horizontal flight speed limit; the instantaneous wind speed of the local micro-meteorology of the flight segment must not exceed the maximum wind resistance limit of the entire aircraft; and the vertical airflow and propulsion power state during the descent phase must be within the effective applicable range of the multi-rotor propulsion power calculation model.

[0092] Understandably, a closed-loop inspection route refers to a sequence of flight paths that start from the UAV base (take-off / landing pad) as the spatial geometric starting point, sequentially execute the entry trajectory of the first photovoltaic array, the in-array inspection trajectory of all designated photovoltaic arrays under the optimal selected flight direction, the lowest energy consumption channel trajectory of seamless topological splicing between adjacent photovoltaic arrays in physical space, and after traversing and covering all designated photovoltaic arrays within the power station once, finally fly back to the original base (or designated recovery base) from the exit of the last photovoltaic array to land. This sequence is a complete and integrated flight path sequence that is strictly connected end to end in physical three-dimensional geographic space, possessing physical continuity and mission closure.

[0093] Specifically, for the candidate access order in the permutation variables, a two-state dynamic programming recursive decision is made on the inspection direction of each photovoltaic row, including: Each decision stage of dynamic programming corresponds to a photovoltaic row, and the stage includes forward inspection status and reverse inspection status. The cumulative cost of the current photovoltaic array in the specified direction is set as the sum of the inspection energy consumption of the photovoltaic array in that direction and the minimum cumulative connection cost of transferring the forward and reverse inspection states of the previous photovoltaic array to the current direction. The connection cost from the base to the entrance of the first photovoltaic row is introduced, and the connection cost from the exit of the last photovoltaic row back to the base, as well as the take-off and landing energy consumption, are introduced. The inspection direction combination under this access sequence is determined by reverse backtracking.

[0094] Furthermore, under the conditions of satisfying the preset constraints on available battery energy, hard exclusion zone obstacles, and UAV flight performance, the output is a closed inspection route that includes the photovoltaic row access sequence, the inspection direction of each photovoltaic row, and the lowest energy consumption channel between photovoltaic rows, including: The photovoltaic row access order is generated by multi-starting-point iterative local search. The initial access order of the multi-starting-point is composed of block-level serpentine order, nearest neighbor order and random sequence. During the iterative search process, the access order is updated sequentially using single-row relocation neighborhood, two-row swap neighborhood, and interval inversion neighborhood, and continuous photovoltaic block relocation is performed during the perturbation phase. For each generated candidate access sequence, a two-state dynamic programming method is called to calculate the total energy consumption of the inspection task. The total energy consumption of the inspection task is verified to be no more than the available energy after deducting the safety reserve from the total battery energy. The inspection scheme with the minimum total energy consumption is selected to obtain the closed inspection route.

[0095] For example, for each generated candidate access order, a two-state dynamic programming method is called to calculate the total energy consumption of the inspection task. The formula for calculating the total energy consumption of the inspection task is as follows: ; in, This indicates the total energy consumption of a complete inspection task. Indicates energy consumption during takeoff and landing. Energy consumption for the connection from the drone base to the first directed inspection status entry point. The connection energy consumption for returning to the drone base from the last directed inspection status exit. This indicates the energy consumption for internal inspections. This indicates the energy consumption of the inter-row connection. For the order of access, For inspection direction, For connecting channels.

[0096] For example, the inspection scheme with the minimum total energy consumption is selected to obtain the closed inspection route. The formula for calculating the minimum total energy consumption is as follows: ; in, Indicates completion photovoltaic row and first Choose direction Minimum total energy at time For positive directional inspection status and This is a reverse directional inspection state. This represents the energy connection cost between two directed inspection states. This indicates the energy consumption of the current directed inspection.

[0097] It is understood that in the embodiments of the present invention, p, q∈{+,-}, p represents the inspection direction selected by the previous photovoltaic row, q represents the inspection direction selected by the current photovoltaic row, "+" represents the forward directional inspection state, and "-" represents the reverse directional inspection state.

[0098] Understandably, multi-starting-point iterative local search refers to a metaheuristic global optimization algorithm architecture that combines multiple initial solution injection, systematic neighborhood local search, and random perturbation exit mechanism. In this algorithm, multiple high-quality heuristic initial solutions and random solutions with significant topological differences are pre-constructed as independent multi-restarting points. Deep local optimization and perturbation iterative loops are performed from different starting points to effectively resist the defect of traditional single heuristic algorithms that are prone to getting trapped in a local minimum. This significantly improves the search coverage and convergence robustness in finding a globally better access order in the photovoltaic row permutation space.

[0099] Understandably, the block-level serpentine sequence refers to a macroscopic iterative heuristic initial access sequence constructed based on the prior spatial topological regularity that physical photovoltaic arrays in photovoltaic power plants are often arranged in parallel rows according to arrays, sub-arrays, or specific geometric blocks. In geometric construction, the power plant is divided into several photovoltaic array blocks from top to bottom or along a specific direction. The UAV traverses adjacent parallel rows within a block in a zigzag pattern, and advances sequentially between blocks according to spatial topological proximity. This initial sequence closely matches the routine manual or automated inspection habits on engineering sites, and can provide a benchmark initial solution with a relatively compact geometric flight path and minimal crossovers for outer layer optimization.

[0100] Understandably, the nearest neighbor order refers to a heuristic initial access sequence constructed using a greedy criterion. This method starts with an initial photovoltaic (PV) row (e.g., the PV row closest to the drone's takeoff base). From the entire set of unvisited candidate PV rows, it uses Euclidean geometric distance or a pre-computed transfer cost lookup table to progressively search for and greedily select the next PV row with the closest geometric spatial distance to the current row outlet as the next hop target, until all designated PV rows within the power plant have been added to the sequence exactly once. This initial solution exhibits extremely high spatial clustering and short-distance transfer characteristics between locally adjacent rows.

[0101] Understandably, a random sequence refers to a discrete sequence generated by uniformly randomizing and unbiasedly shuffling the order of all physical photovoltaic arrays to be inspected within a power plant using a pseudo-random number generator, based on a set of discrete permutations consisting of unique identifiers. Random sequences do not rely on any prior spatial geometric knowledge and aim to introduce completely random topological diversity into a multi-starting-point search pool. This expands the algorithm's exploration range in different unknown basins within the overall solution space and prevents the omission of potential globally optimal solution types due to prior greedy heuristic biases.

[0102] Understandably, single-row relocation neighborhood refers to a one-dimensional permutation operator and its derived set of candidate solutions that fine-tune and improve the candidate access order during the local search phase. This operation extracts any selected single-row photovoltaic element from the current access order, removes it from its original position, and attempts to re-insert it into any other specified position in the permutation (insertion into a gap or before / after insertion), while maintaining the relative order of the remaining photovoltaic rows. This operator is mainly used to prune long-distance cross-turning routes and correct the "flyback" phenomenon caused by a local permutation order reversal.

[0103] Understandably, the two-row swap neighborhood refers to a pairwise permutation operator and its derived candidate solution set that topologically perturbs and improves the candidate access order during the local search phase. This operation arbitrarily specifies two different photovoltaic rows in the current access order and directly swaps their position indices in the sequence, while completely preserving the positions of the remaining photovoltaic rows in the sequence. This operator can break suboptimal pairwise binding relationships between rows with relatively small perturbations, promoting transitional deconstruction and topological reorganization between sub-arrays or between different rows of the same array.

[0104] Understandably, interval reversal neighborhood refers to the combinatorial optimization operator and its derived set of candidate solutions that perform topological mirror reversal on a continuous subsequence in the candidate access order during the local search phase. This operation randomly or systematically extracts a continuous photovoltaic subsequence interval defined by any two dividing points from the current access order. The order of visits to all photovoltaic arrays within the closed interval is reversed (i.e., the order of visits in the original sequence is reversed). To the Reversing the order to the next To the (The remaining intervals remain unchanged). At the geometric level, this operation has a powerful ability to untangle spatial crossings, which can significantly eliminate ineffective detours and energy waste caused by route "knotting" in large-scale transitions.

[0105] Understandably, two-state dynamic programming refers to an inner-layer deterministic recursive algorithm that, given a fixed candidate access order for an outer photovoltaic (PV) row, precisely solves for the optimal combination of internal inspection directions for each PV row using the properties of optimal substructure. In this algorithm: each stage strictly corresponds to a physical PV row in the access order; the state space is a binary discrete state, explicitly defined as the forward inspection state (state 0, flying from the inlet to the outlet along a preset forward direction) and the reverse inspection state (state 1, flying from the outlet to the inlet along a preset reverse direction) of the UAV flying within the row; the state transition cost includes the energy consumption of the current row's internal inspection in the corresponding direction, and the minimum energy consumption channel connection cost between rows corresponding to the transition from the two optional exit states of the previous row to the current row's inlet state; by recursively accumulating and minimizing the accumulated energy consumption cost from the first stage (including base entry energy consumption) to subsequent stages and executing a backtracking mark in the final stage (including return to base energy consumption), it can achieve extremely high computational efficiency (polynomial time complexity). )exist In a vast array of combinations, we can rigorously and deterministically determine the combination of inspection directions that maximizes the total energy consumption of the closed-loop inspection for the entire task, along with the corresponding total energy consumption.

[0106] This invention acquires the digital elevation model, photovoltaic (PV) array, hard exclusion zone, safety buffer zone, spatial wind vector field, and camera parameters of the photovoltaic power station to be inspected. Based on the tilt angle of the PV modules in the PV array and the camera imaging geometric parameters in the camera parameters, it constructs the spatial inspection trajectory and upper limit of the inspection speed for each PV array. This provides comprehensive geometric trajectory input and high-quality geometric configuration data. Furthermore, for each PV array, it constructs forward and reverse directional inspection states with spatial geometric overlap but opposite flight directions. Using the flight segment's spatial location and flight direction as indices, it determines the absolute altitude of the midpoint of the flight segment using the digital elevation model, calculates the local air density at that absolute altitude based on a standard atmospheric model, and combines this with... The spatial wind vector field determines the horizontal relative air velocity and vertical velocity components of the flight segment, which can improve the accuracy of energy consumption calculations caused by environmental anisotropy. Furthermore, the free space defined by the digital elevation model, hard exclusion zones, and safety buffer zones is discretized into a directed grid map. The height of each grid node is set based on the corresponding location's digital elevation model elevation and a preset transition ground clearance. Each feasible directed edge in the directed grid map is assigned a three-dimensional spatial geometric distance weight, and the flight segment energy weight is dynamically updated based on absolute altitude, spatial wind vector field, and a standard atmospheric model. Infeasible directed edges in the directed grid map are eliminated based on preset UAV flight performance constraints, preventing safety hazards such as flight paths crossing obstacles or inducing aerodynamic stall. In addition, regarding... Between the UAV base and each directed inspection state, and between any two directed inspection states of different photovoltaic arrays, graph searches are performed in the same directed grid graph based on three-dimensional spatial geometric distance weights and segment energy weights, respectively. The shortest distance channel and the lowest energy consumption channel are obtained and retained. The range and energy consumption of each type of channel are recorded, and four sets of directed connection matrices are generated representing the range of the shortest distance channel, the energy consumption along the shortest distance channel, the energy consumption of the lowest energy consumption channel, and the range of the lowest energy consumption channel. This allows for the parallel retention of both shortest distance and lowest energy consumption channels between the UAV base and each inspection state, and between directed inspection states of different photovoltaic arrays, generating four sets of directed connection matrices containing range and energy consumption information, thereby supporting... Subsequently, energy-priority and range-measurable connection decisions are made under asymmetric directed costs. Finally, the access order of the photovoltaic rows traversed from the outer layer is used as the permutation variable. For the candidate access order in the permutation variable, two-state dynamic programming is used to recursively decide the inspection direction of each photovoltaic row. The lowest energy consumption channel between adjacent directed inspection states is extracted from the four sets of directed connection matrices. Under the conditions of satisfying the preset battery available energy constraints, hard exclusion zone obstacle constraints, and UAV flight performance constraints, a closed inspection route containing the photovoltaic row access order, the inspection direction of each photovoltaic row, and the lowest energy consumption channel between photovoltaic rows is output. This can improve the accuracy and efficiency of environmentally adaptive multi-channel energy sensing path planning for UAV inspection of high-altitude photovoltaic power stations.

[0107] like Figure 2The diagram shown is a schematic of the segment energy weight acquisition process of the environmental adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants provided by an embodiment of the present invention.

[0108] like Figure 3 The diagram shown is a schematic of the shortest path distance for an environment-adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants, provided by an embodiment of the present invention.

[0109] like Figure 4 The diagram shown is a schematic of the lowest energy consumption channel of an environmental adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants provided by an embodiment of the present invention.

[0110] like Figure 5 The diagram shown is a functional block diagram of an environmentally adaptive multi-channel energy sensing path planning device for UAV inspection of high-altitude photovoltaic power plants, provided by an embodiment of the present invention.

[0111] The environmental adaptive multi-channel energy sensing path planning device 100 for UAV inspection of high-altitude photovoltaic power plants described in this invention can be installed in an electronic device. Depending on the functions implemented, the environmental adaptive multi-channel energy sensing path planning device 100 for UAV inspection of high-altitude photovoltaic power plants may include an environmental sensing and inspection state modeling module 101, a directed graph construction and multi-channel connection matrix solution module 102, and a global flight path joint planning and output module 103.

[0112] The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0113] In this embodiment, the functions of each module / unit are as follows: The environmental perception and inspection status modeling module 101 is used to acquire the digital elevation model, photovoltaic arrangement, hard exclusion zone, safety buffer zone, spatial wind vector field and camera parameters of the photovoltaic power station to be inspected, and construct the spatial inspection trajectory and inspection speed limit corresponding to each photovoltaic row based on the tilt angle of the photovoltaic modules in the photovoltaic arrangement and the camera imaging geometric parameters in the camera parameters. For each photovoltaic row in the photovoltaic arrangement, a forward directional inspection state and a reverse directional inspection state with spatial geometric overlap but opposite flight directions are constructed respectively. Based on the spatial position and flight direction of the flight segment as an index, the absolute altitude of the midpoint of the flight segment is determined using the digital elevation model, the local air density at the absolute altitude is calculated based on the standard atmospheric model, and the horizontal relative air velocity and vertical velocity components of the flight segment are determined by combining the spatial wind vector field.

[0114] The directed graph construction and multi-channel connection matrix solution module 102 is used to discretize the free space defined by the digital elevation model, hard exclusion zone and safety buffer into a directed grid graph. The height of each grid node is set based on the digital elevation model elevation of the corresponding location and the preset transfer ground clearance. Each feasible directed edge in the directed grid graph is assigned a three-dimensional spatial geometric distance weight, and the segment energy weight is dynamically updated according to the absolute altitude, spatial wind vector field and standard atmospheric model. Infeasible directed edges in the directed grid graph are eliminated according to the preset UAV flight performance constraints. For the UAV base and each directed inspection state, and between any two different photovoltaic row directed inspection states, a graph search is performed in the same directed grid graph based on the three-dimensional spatial geometric distance weight and the segment energy weight, respectively. The shortest distance channel and the lowest energy consumption channel are obtained and retained, and the flight distance and channel energy consumption of each type of channel are recorded. Four sets of directed connection matrices are generated to represent the shortest distance channel flight distance, the energy consumption along the shortest distance channel, the energy consumption of the lowest energy consumption channel, and the flight distance of the lowest energy consumption channel.

[0115] The global route joint planning and output module 103 is used as the photovoltaic row access order of the outer traversal as the arrangement variable. For the candidate access order in the arrangement variable, it calls two-state dynamic programming to recursively decide the inspection direction of each photovoltaic row. It also extracts the lowest energy consumption channel between adjacent directed inspection states from four sets of directed connection matrices. Under the conditions of satisfying the preset battery available energy constraints, hard exclusion zone obstacle constraints and UAV flight performance constraints, it outputs a closed inspection route that includes the photovoltaic row access order, the inspection direction of each photovoltaic row, and the lowest energy consumption channel between photovoltaic rows.

[0116] like Figure 6 The diagram shown is a structural schematic of an electronic device for implementing an environmentally adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants, according to an embodiment of the present invention.

[0117] The electronic device may include a processor 10, a memory 11, a communication bus 12, and a communication interface 13. It may also include a computer program stored in the memory 11 and capable of running on the processor 10, such as an environment-adaptive multi-channel energy sensing path planning method program for UAV inspection of high-altitude photovoltaic power plants.

[0118] In some embodiments, the processor 10 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 10 is the control unit of the electronic device, connecting various components of the entire electronic device through various interfaces and lines. It executes programs or modules stored in the memory 11 (e.g., executing an environmentally adaptive multi-channel energy sensing path planning method program for UAV inspection of high-altitude photovoltaic power stations), and calls data stored in the memory 11 to perform various functions of the electronic device and process data.

[0119] The memory 11 includes at least one type of readable storage medium, including flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of an electronic device, such as a portable hard drive. In other embodiments, the memory 11 can be an external storage device of the electronic device, such as a plug-in portable hard drive, Smart Media Card (SMC), Secure Digital (SD) card, Flash Card, etc. Furthermore, the memory 11 can include both internal and external storage units of the electronic device. The memory 11 can be used not only to store application software and various types of data installed on the electronic device, such as the code of an environmentally adaptive multi-channel energy sensing path planning method program for UAV inspection of high-altitude photovoltaic power stations, but also to temporarily store data that has been output or will be output.

[0120] The communication bus 12 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. The bus is configured to enable communication between the memory 11 and at least one processor 10, etc.

[0121] The communication interface 13 is used for communication between the aforementioned electronic device and other devices, including a network interface and a user interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a Wi-Fi interface, Bluetooth interface, etc.), typically used to establish communication connections between the electronic device and other electronic devices. The user interface may be a display, an input unit (such as a keyboard), or optionally, a standard wired or wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the electronic device and to display a visual user interface.

[0122] Figure 6 Only electronic devices with components are shown; it will be understood by those skilled in the art that... Figure 6 The structure shown does not constitute a limitation on the electronic device and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0123] For example, although not shown, the electronic device may also include a power supply (such as a battery) to power the various components. Preferably, the power supply can be logically connected to the at least one processor 10 through a power management device, thereby enabling functions such as charging management, discharging management, and power consumption management. The power supply may also include one or more DC or AC power supplies, recharging devices, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0124] It should be understood that the embodiments described are for illustrative purposes only and are not limited to this structure in the scope of the patent application.

[0125] The program for an environment-adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants, stored in the memory 11 of the electronic device, is a combination of multiple instructions. When run in the processor 10, it can achieve the following: The system acquires the digital elevation model, photovoltaic layout, hard exclusion zone, safety buffer zone, spatial wind vector field, and camera parameters of the photovoltaic power station to be inspected. Based on the tilt angle of the photovoltaic modules in the photovoltaic layout and the camera imaging geometric parameters in the camera parameters, the system constructs the spatial inspection trajectory and inspection speed limit for each photovoltaic row. For each photovoltaic array, a forward directional inspection state and a reverse directional inspection state with spatial geometric overlap but opposite flight directions are constructed. Based on the spatial position and flight direction of the flight segment, the absolute altitude of the midpoint of the flight segment is determined using a digital elevation model. The local air density at the absolute altitude is calculated based on a standard atmospheric model. The horizontal relative air velocity and vertical velocity components of the flight segment are determined by combining the spatial wind vector field. The free space defined by the digital elevation model, hard exclusion zone and safety buffer is discretized into a directed grid graph. The height of each grid node is set based on the digital elevation model elevation of the corresponding location and the preset transfer ground clearance. Each feasible directed edge in the directed grid graph is assigned a three-dimensional spatial geometric distance weight. The flight segment energy weight is dynamically updated according to the absolute altitude, spatial wind vector field and standard atmospheric model. Infeasible directed edges in the directed grid graph are removed according to the preset UAV flight performance constraints. For the relationship between the UAV base and each directed inspection state, as well as between any two different photovoltaic arrays, graph search is performed in the same directed grid graph based on three-dimensional spatial geometric distance weight and segment energy weight, respectively. The shortest distance channel and the lowest energy consumption channel are obtained and retained, and the range and channel energy consumption of the two types of channels are recorded. Four sets of directed connection matrices are generated to represent the range of the shortest distance channel, the energy consumption of flying along the shortest distance channel, the energy consumption of the lowest energy consumption channel, and the range of the lowest energy consumption channel. Using the access order of the photovoltaic rows in the outer traversal as the permutation variable, a two-state dynamic programming method is called to recursively decide the inspection direction of each photovoltaic row for the candidate access order in the permutation variable. The lowest energy consumption channel between adjacent directed inspection states is extracted from the four sets of directed connection matrices. Under the conditions of satisfying the preset battery available energy constraints, hard exclusion zone obstacle constraints and UAV flight performance constraints, the closed inspection route containing the access order of photovoltaic rows, the inspection direction of each photovoltaic row and the lowest energy consumption channel between photovoltaic rows is output.

[0126] Specifically, the specific implementation method of the processor 10 for the above instructions can be referred to the description of the relevant steps in the corresponding embodiment of the accompanying drawings, and will not be repeated here.

[0127] Furthermore, if the modules / units integrated in the electronic device 1 are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0128] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor of an electronic device, can perform the following: The system acquires the digital elevation model, photovoltaic layout, hard exclusion zone, safety buffer zone, spatial wind vector field, and camera parameters of the photovoltaic power station to be inspected. Based on the tilt angle of the photovoltaic modules in the photovoltaic layout and the camera imaging geometric parameters in the camera parameters, the system constructs the spatial inspection trajectory and inspection speed limit for each photovoltaic row. For each photovoltaic array, a forward directional inspection state and a reverse directional inspection state with spatial geometric overlap but opposite flight directions are constructed. Based on the spatial position and flight direction of the flight segment, the absolute altitude of the midpoint of the flight segment is determined using a digital elevation model. The local air density at the absolute altitude is calculated based on a standard atmospheric model. The horizontal relative air velocity and vertical velocity components of the flight segment are determined by combining the spatial wind vector field. The free space defined by the digital elevation model, hard exclusion zone and safety buffer is discretized into a directed grid graph. The height of each grid node is set based on the digital elevation model elevation of the corresponding location and the preset transfer ground clearance. Each feasible directed edge in the directed grid graph is assigned a three-dimensional spatial geometric distance weight. The flight segment energy weight is dynamically updated according to the absolute altitude, spatial wind vector field and standard atmospheric model. Infeasible directed edges in the directed grid graph are removed according to the preset UAV flight performance constraints. For the relationship between the UAV base and each directed inspection state, as well as between any two different photovoltaic arrays, graph search is performed in the same directed grid graph based on three-dimensional spatial geometric distance weight and segment energy weight, respectively. The shortest distance channel and the lowest energy consumption channel are obtained and retained, and the range and channel energy consumption of the two types of channels are recorded. Four sets of directed connection matrices are generated to represent the range of the shortest distance channel, the energy consumption of flying along the shortest distance channel, the energy consumption of the lowest energy consumption channel, and the range of the lowest energy consumption channel. Using the access order of the photovoltaic rows in the outer traversal as the permutation variable, a two-state dynamic programming method is called to recursively decide the inspection direction of each photovoltaic row for the candidate access order in the permutation variable. The lowest energy consumption channel between adjacent directed inspection states is extracted from the four sets of directed connection matrices. Under the conditions of satisfying the preset battery available energy constraints, hard exclusion zone obstacle constraints and UAV flight performance constraints, the closed inspection route containing the access order of photovoltaic rows, the inspection direction of each photovoltaic row and the lowest energy consumption channel between photovoltaic rows is output.

[0129] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0130] The modules described as separate components may or may not be physically separate. The components shown as modules 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 the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0131] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0132] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0133] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the invention. No appended diagram markings in the claims should be construed as limiting the scope of the claims.

[0134] The blockchain referred to in this invention is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Essentially, a blockchain is a decentralized database, a chain of data blocks linked together using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain can include an underlying blockchain platform, a platform product service layer, and an application service layer.

[0135] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0136] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices recited in a system claim may also be implemented by a single unit or device through software or hardware. The terms "first," "second," etc., are used to indicate names and do not indicate any specific order.

[0137] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An environment-adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants, characterized in that, The method includes: The system acquires the digital elevation model, photovoltaic layout, hard exclusion zone, safety buffer zone, spatial wind vector field, and camera parameters of the photovoltaic power station to be inspected. Based on the tilt angle of the photovoltaic modules in the photovoltaic layout and the camera imaging geometric parameters in the camera parameters, the system constructs the spatial inspection trajectory and inspection speed limit for each photovoltaic row. For each photovoltaic array, a forward directional inspection state and a reverse directional inspection state with spatial geometric overlap but opposite flight directions are constructed. Based on the spatial position and flight direction of the flight segment, the absolute altitude of the midpoint of the flight segment is determined using a digital elevation model. The local air density at the absolute altitude is calculated based on a standard atmospheric model. The horizontal relative air velocity and vertical velocity components of the flight segment are determined by combining the spatial wind vector field. The free space defined by the digital elevation model, hard exclusion zone and safety buffer is discretized into a directed grid graph. The height of each grid node is set based on the digital elevation model elevation of the corresponding location and the preset transfer ground clearance. Each feasible directed edge in the directed grid graph is assigned a three-dimensional spatial geometric distance weight. The flight segment energy weight is dynamically updated according to the absolute altitude, spatial wind vector field and standard atmospheric model. Infeasible directed edges in the directed grid graph are removed according to the preset UAV flight performance constraints. For the relationship between the UAV base and each directed inspection state, as well as between any two different photovoltaic arrays, graph search is performed in the same directed grid graph based on three-dimensional spatial geometric distance weight and segment energy weight, respectively. The shortest distance channel and the lowest energy consumption channel are obtained and retained, and the range and channel energy consumption of the two types of channels are recorded. Four sets of directed connection matrices are generated to represent the range of the shortest distance channel, the energy consumption of flying along the shortest distance channel, the energy consumption of the lowest energy consumption channel, and the range of the lowest energy consumption channel. Using the access order of the photovoltaic rows in the outer traversal as the permutation variable, a two-state dynamic programming method is called to recursively decide the inspection direction of each photovoltaic row for the candidate access order in the permutation variable. The lowest energy consumption channel between adjacent directed inspection states is extracted from the four sets of directed connection matrices. Under the conditions of satisfying the preset battery available energy constraints, hard exclusion zone obstacle constraints and UAV flight performance constraints, the closed inspection route containing the access order of photovoltaic rows, the inspection direction of each photovoltaic row and the lowest energy consumption channel between photovoltaic rows is output.

2. The environmentally adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants as described in claim 1, characterized in that, The process of constructing the spatial inspection trajectory and upper limit of inspection speed for each photovoltaic row based on the tilt angle of the photovoltaic modules in the photovoltaic array and the camera imaging geometric parameters in the camera parameters includes: The center line of the photovoltaic array is translated along the normal direction of the photovoltaic module surface by a preset camera normal distance to generate a spatial inspection trajectory parallel to the photovoltaic module surface; The maximum inspection speed is calculated based on the camera normal distance, sensor physical dimensions along the flight path, camera focal length, camera minimum shooting interval, and forward overlap rate. The actual inspection speed is then limited to no more than the maximum inspection speed.

3. The environmentally adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants as described in claim 1, characterized in that... The dynamic updating of flight segment energy weights based on absolute altitude, spatial wind vector field, and standard atmospheric model includes: The spatial trajectory is discretized into flight segment units. The absolute altitude of the midpoint of the flight segment unit is input into the standard atmospheric model. The local temperature is calculated from the standard sea level temperature and the temperature lapse rate. The local air pressure is calculated from the standard sea level pressure, local temperature, gravitational acceleration, and air gas constant. Then the local air density is calculated. The ideal induction velocity for hovering is calculated from the local air density, and the profile power reference is scaled. The horizontal wind vector at the midpoint of the flight segment is sampled from the spatial wind vector field. The horizontal relative air velocity is determined by the difference between the UAV's horizontal ground speed vector and the horizontal wind vector. The horizontal flight power component is then calculated. Calculate the vertical motion power increment based on the vertical velocity components of the flight segment; For turning segments with a preset finite turning radius, the thrust multiple is determined based on the horizontal relative air speed and the preset turning radius, and the additional turning power is calculated. For straight grid edges in the directed grid diagram, the additional turning power is set to zero. The instantaneous total power is obtained by adding the horizontal flight power component, vertical motion power increment, turning additional power, fixed power consumption of avionics and communication, and camera power consumption. The directional energy weight that varies with the spatial position and flight direction of the flight segment is generated based on the instantaneous total power and the flight time integral of the segment.

4. The environmentally adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants as described in claim 1, characterized in that... Discretizing the free space defined by the digital elevation model, hard exclusion zone, and safety buffer into a directed raster graph includes: The directed grid graph is set to an eight-neighbor topological connection, with both endpoints of the connection edge located in free space, and the two orthogonal adjacent nodes corresponding to the diagonal connection edge located in free space to prevent corner points from crossing. The flight time and velocity components in each direction are determined based on the three-dimensional geometric length of the connecting edges and the set transfer speed. Directed edges that violate the maximum climb speed limit, maximum descent speed limit, maximum horizontal flight speed limit, local maximum wind speed limit, or the applicable range limit of the descent propulsion model are eliminated.

5. The environmentally adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants as described in claim 1, characterized in that, The graph search is performed on the same directed grid graph based on three-dimensional spatial geometric distance weight and segment energy weight, respectively, to obtain and retain the shortest distance channel and the lowest energy consumption channel. The range and energy consumption of each type of channel are recorded, and four sets of directed connectivity matrices are generated representing the range of the shortest distance channel, the energy consumption along the shortest distance channel, the energy consumption of the lowest energy consumption channel, and the range of the lowest energy consumption channel. For each exit node of a directed inspection state, a single-source shortest path search is performed with three-dimensional spatial geometric distance weight and segment energy weight respectively, to obtain the shortest distance channel and the lowest energy consumption channel to each target directed inspection state entry node respectively. The search results are stored in the corresponding directed connection matrices, and the two types of channels are associated with the predecessor node or the predecessor edge sequence, respectively. When the shortest path and the lowest energy consumption path between the same state pair coincide in spatial geometry, they share the same path sequence and record the corresponding travel cost and energy consumption cost respectively.

6. The environmentally adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants as described in claim 1, characterized in that, The step of using two-state dynamic programming to recursively decide the inspection direction of each photovoltaic row based on the candidate access order in the permutation variables includes: Each decision stage of dynamic programming corresponds to a photovoltaic row, and the stage includes forward inspection status and reverse inspection status. The cumulative cost of the current photovoltaic array in the specified direction is set as the sum of the inspection energy consumption of the photovoltaic array in that direction and the minimum cumulative connection cost of transferring the forward and reverse inspection states of the previous photovoltaic array to the current direction. The connection cost from the base to the entrance of the first photovoltaic row is introduced, and the connection cost from the exit of the last photovoltaic row back to the base, as well as the take-off and landing energy consumption, are introduced. The inspection direction combination under this access sequence is determined by reverse backtracking.

7. The environmentally adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants as described in claim 1, characterized in that, Under the conditions of satisfying preset constraints on available battery energy, hard obstacle removal zone, and UAV flight performance, the output is a closed inspection route that includes the photovoltaic array access sequence, the inspection direction of each photovoltaic array, and the lowest energy consumption channel between photovoltaic arrays, including: The photovoltaic row access order is generated by multi-starting point iterative local search. The initial access order of the multi-starting point consists of block-level serpentine order, nearest neighbor order and random sequence. During the iterative search process, the access order is updated sequentially using single-row relocation neighborhood, two-row swap neighborhood, and interval inversion neighborhood, and continuous photovoltaic block relocation is performed during the perturbation phase. For each generated candidate access sequence, a two-state dynamic programming method is called to calculate the total energy consumption of the inspection task. The total energy consumption of the inspection task is verified to be no more than the available energy after deducting the safety reserve from the total battery energy. The inspection scheme with the minimum total energy consumption is selected to obtain the closed inspection route.

8. An environment-adaptive multi-channel energy sensing path planning device for UAV inspection of high-altitude photovoltaic power plants, characterized in that, The device can implement the environmentally adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants as described in any one of claims 1 to 7, and the device includes: The environmental perception and inspection status modeling module is used to acquire the digital elevation model, photovoltaic layout, hard exclusion zone, safety buffer zone, spatial wind vector field, and camera parameters of the photovoltaic power station to be inspected. Based on the tilt angle of the photovoltaic modules in the photovoltaic layout and the camera imaging geometric parameters in the camera parameters, it constructs the spatial inspection trajectory and inspection speed limit corresponding to each photovoltaic row. For each photovoltaic row in the photovoltaic layout, it constructs forward directional inspection status and reverse directional inspection status with spatial geometric overlap but opposite flight directions. Based on the spatial position and flight direction of the flight segment as indexes, it uses the digital elevation model to determine the absolute altitude of the midpoint of the flight segment, calculates the local air density at the absolute altitude based on the standard atmospheric model, and determines the horizontal relative air velocity and vertical velocity components of the flight segment by combining the spatial wind vector field. The directed graph construction and multi-channel connection matrix solution module is used to discretize the free space defined by the digital elevation model, hard exclusion zone, and safety buffer into a directed grid graph. The height of each grid node is set based on the digital elevation model elevation of the corresponding location and the preset transfer ground clearance. Each feasible directed edge in the directed grid graph is assigned a three-dimensional spatial geometric distance weight, and the segment energy weight is dynamically updated according to the absolute altitude, spatial wind vector field, and standard atmospheric model. Infeasible directed edges in the directed grid graph are eliminated according to the preset UAV flight performance constraints. For the UAV base and each directed inspection state, as well as between any two different photovoltaic row units, a graph search is performed in the same directed grid graph based on the three-dimensional spatial geometric distance weight and the segment energy weight, respectively. The shortest distance channel and the lowest energy consumption channel are obtained and retained, respectively. The flight distance and channel energy consumption of the two types of channels are recorded, and four sets of directed connection matrices are generated to represent the shortest distance channel flight distance, the energy consumption along the shortest distance channel, the energy consumption of the lowest energy consumption channel, and the flight distance of the lowest energy consumption channel. The global route joint planning and output module uses the photovoltaic row access order of the outer traversal as the arrangement variable. For the candidate access order in the arrangement variable, it calls two-state dynamic programming to recursively decide the inspection direction of each photovoltaic row. It also extracts the lowest energy consumption channel between adjacent directed inspection states from four sets of directed connection matrices. Under the conditions of satisfying the preset battery available energy constraints, hard exclusion zone obstacle constraints and UAV flight performance constraints, it outputs a closed inspection route that includes the photovoltaic row access order, the inspection direction of each photovoltaic row, and the lowest energy consumption channel between photovoltaic rows.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to execute the environmental adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the environmentally adaptive multi-channel energy sensing path planning method for UAV inspection of high-altitude photovoltaic power plants as described in any one of claims 1 to 7.