Route planning method, device and storage medium

Abstract

The application discloses a route planning method and device and a storage medium, relates to the technical field of data processing, and discloses a route planning method, which comprises the following steps: determining the geometric characteristics of a fan blade; dividing the fan blade into different partitions based on the geometric characteristics, wherein each partition corresponds to a different part of the fan blade; determining a route planning strategy suitable for each partition, and performing route planning on the corresponding partition based on the route planning strategy to obtain a route corresponding to each partition; and splicing the routes corresponding to each partition to obtain a target route of a UAV, wherein the target route is a flight path of the UAV when the UAV inspects the fan blade. According to the technical scheme, the fan blade can be divided into different partitions according to the geometric characteristics of different parts of the fan blade, and the route corresponding to each partition can be determined accordingly, so that the accuracy of the target route of the UAV is improved.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to route planning methods, equipment and storage media. Background Technology

[0002] As a key component of wind turbine units, wind turbine blades are exposed to harsh natural environments for extended periods, enduring alternating loads, wind and sand erosion, lightning strikes, and ultraviolet radiation. This makes them prone to defects such as cracks, corrosion, gel coat peeling, and leading-edge wear. Failure to detect and address these defects in a timely manner can lead to serious accidents such as blade breakage, resulting in significant economic losses and safety risks. Therefore, regular inspections of wind turbine blades are crucial.

[0003] Currently, before drones inspect wind turbine blades, the system imports a 3D model of the wind turbine blades, sets globally unified relative flight distances, unified waypoint spacing, and fixed camera pitch angles, and generates a terrain-following parallel flight path from the blade root to the blade tip, enabling drones to inspect wind turbine blades based on this terrain-following parallel flight path.

[0004] However, wind turbine blades are typical complex non-standard geometries with long dimensions, variable cross-sections, and variable curvatures. Using a globally unified static flight path planning algorithm, the surface of the blade root region is gentle, and the fixed waypoint step size will generate a large number of dense redundant waypoints, consuming ineffective endurance; while in regions where the chord length shrinks sharply, such as the blade tip, the original step size will cause the overlap rate to drop precipitously; due to uneven spatial sampling, data redundancy and local omissions coexist, resulting in low accuracy of the final target flight path generated by the UAV. Summary of the Invention

[0005] The main purpose of this application is to provide a flight path planning method, device and storage medium, which aims to divide wind turbine blades into different zones and determine the flight path corresponding to each zone in a targeted manner, so as to improve the accuracy of the target flight path of the UAV.

[0006] To achieve the above objectives, this application proposes a route planning method, comprising: Determine the geometric characteristics of the wind turbine blades; Based on geometric features, the wind turbine blades are divided into different zones, with each zone corresponding to a different part of the wind turbine blade; Determine the appropriate route planning strategy for each partition, and perform route planning for the corresponding partitions based on the route planning strategy to obtain the routes corresponding to each partition. By piecing together the flight paths corresponding to each zone, the target flight path corresponding to the UAV is obtained. The target flight path is the flight path of the UAV when inspecting the wind turbine blades.

[0007] In addition, to achieve the above objectives, this application also proposes a route planning device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the route planning method as described above.

[0008] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the route planning method described above.

[0009] The present application proposes one or more technical solutions, which have at least the following technical effects: The present application first determines the geometric characteristics of the wind turbine blades, then divides the blades into multiple partitions corresponding to different parts based on the geometric characteristics, matches an appropriate route planning strategy to each partition and generates a route independently, and finally splices the routes of each partition to obtain the target route of the UAV; This solution effectively solves the problem of uneven spatial sampling caused by a globally unified route, can adaptively match the geometric characteristics of the wind turbine blades, realize the differentiation of different regions and generate refined routes, and improve the accuracy of the generated routes. Attached Figure Description

[0010] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0011] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 This is a flowchart illustrating an embodiment of the route planning method of this application. Figure 2 This is a detailed flowchart of step S10 in an embodiment of the route planning method of this application; Figure 3 This is a detailed flowchart of step S40 in an embodiment of the route planning method of this application; Figure 4 This is a structural schematic diagram of the route planning equipment used in this application.

[0013] The purpose, features, and advantages of this application 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 technical solutions of this application and are not intended to limit this application.

[0015] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0016] As a key component of wind turbine units, wind turbine blades are exposed to harsh natural environments for extended periods, enduring alternating loads, wind and sand erosion, lightning strikes, and ultraviolet radiation. This makes them prone to defects such as cracks, corrosion, gel coat peeling, and leading-edge wear. Failure to detect and address these defects in a timely manner can lead to serious accidents such as blade breakage, resulting in significant economic losses and safety risks. Therefore, regular inspections of wind turbine blades are crucial.

[0017] Currently, before drones inspect wind turbine blades, the system imports a 3D model of the wind turbine blades, sets globally unified relative flight distances, unified waypoint spacing, and fixed camera pitch angles, and generates a terrain-following parallel flight path from the blade root to the blade tip, enabling drones to inspect wind turbine blades based on this terrain-following parallel flight path.

[0018] However, wind turbine blades are typical complex non-standard geometries with long dimensions, variable cross-sections, and variable curvatures. Using a globally unified static flight path planning algorithm, the surface of the blade root region is gentle, and the fixed waypoint step size will generate a large number of dense redundant waypoints, consuming ineffective endurance; while in regions where the chord length shrinks sharply, such as the blade tip, the original step size will cause the overlap rate to drop precipitously; due to uneven spatial sampling, data redundancy and local omissions coexist, resulting in low accuracy of the final target flight path generated by the UAV.

[0019] To address the aforementioned issues, this application proposes a flight path planning method. The main technical solutions include: determining the geometric characteristics of the wind turbine blades; dividing the wind turbine blades into different partitions based on the geometric characteristics, wherein each partition corresponds to a different part of the wind turbine blade; determining the flight path planning strategy suitable for each partition, and performing flight path planning on the corresponding partitions based on the flight path planning strategy to obtain the flight path corresponding to each partition; and concatenating the flight paths corresponding to each partition to obtain the target flight path corresponding to the UAV, wherein the target flight path is the flight path of the UAV when inspecting the wind turbine blades.

[0020] This application first determines the geometric characteristics of the wind turbine blades, then divides the blades into multiple partitions corresponding to different parts based on these geometric characteristics. Each partition is matched with an appropriate flight path planning strategy and an independent flight path is generated. Finally, the flight paths of each partition are spliced ​​together to obtain the target flight path of the UAV. This scheme effectively solves the problem of uneven spatial sampling caused by a globally unified flight path, can adaptively match the geometric characteristics of the wind turbine blades, achieve differentiation in different regions, generate refined flight paths, and improve the accuracy of the generated flight paths.

[0021] It should be noted that the executing entity in this embodiment can be a route planning device, etc. This route planning device can be a server, or a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone.

[0022] Based on this, embodiments of this application provide a route planning method, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the route planning method of this application. In this embodiment, the route planning method includes steps S10 to S40: Step S10: Determine the geometric characteristics of the wind turbine blades; Geometric features refer to a set of parameters that quantify the shape, surface morphology, cross-sectional dimensions, and curvature of wind turbine blades. They reflect the differences in shape across different parts of the blade and form the basis for subsequent zoning and flight path planning. Geometric features provide criteria for blade zoning, distinguishing morphological differences across different parts of the blade through quantified parameters. This provides a foundation for adapting each zone to corresponding flight path planning strategies, ensuring that flight path planning aligns with the actual geometric shape of the blade. Specifically, these geometric features include the maximum principal curvature, minimum principal curvature, and Gaussian curvature of the blade's three-dimensional mesh vertices; the maximum chord length of each cross-section along the spanwise direction of the blade; and parameters such as the spanwise principal axis direction, the local normal vectors of each vertex on the blade surface, and the surface curvature radii of different parts of the blade.

[0023] In one alternative approach, a drone can be controlled to fly along the surface of the wind turbine blades at preset step lengths and capture images of the blades at fixed points. These images are then stitched together to obtain a complete image of the wind turbine blades. Image analysis is then performed on this complete image to obtain the geometric features of the wind turbine blades. Specifically, the analysis process can involve using a pre-trained network model to analyze the complete image and obtain the geometric features of the wind turbine blades. This network model can be trained using a large number of image samples labeled with geometric features. Alternatively, a finely tuned large language model combined with prompt text can be used to analyze the complete image and obtain the geometric features of the wind turbine blades.

[0024] In another optional approach, a 3D mesh model of the wind turbine blade can be imported first. Each vertex in the 3D mesh is then used as a target vertex, and a local tangent plane coordinate system is established with the target vertex as the origin. A K-DTree search is used to find the neighboring vertices of the target vertex. The local quadratic surface equation is fitted using the least squares method, the quadratic coefficients are extracted, and a Hessian matrix is ​​constructed. Eigenvalue decomposition is performed on the Hessian matrix to obtain the maximum principal curvature, minimum principal curvature, and Gaussian curvature of each target vertex. Simultaneously, a series of slice planes orthogonal to the spanwise principal axis are generated along the blade's spanwise principal axis at preset step sizes. The intersection of these slice planes with the 3D mesh model of the wind turbine blade is calculated, and the cross-sections at each slice location are extracted. A convex hull algorithm is performed on these cross-sections to calculate the Euclidean distance between all vertex pairs on the cross-section, and the maximum value is determined as the maximum chord length of that cross-section. Finally, the curvatures of all vertices and the maximum chord lengths of all cross-sections are integrated to obtain the geometric characteristics of the wind turbine blade.

[0025] Step S20: Based on geometric features, the wind turbine blades are divided into different partitions, where each partition corresponds to a different part of the wind turbine blade; Zoned designation refers to dividing the entire wind turbine blade into multiple independent regions based on its geometric characteristics. Each region corresponds to a specific part of the blade, and the geometry, curvature, and cross-sectional dimensions of each region may differ. Zoned design enables differentiated and refined flight path planning, avoiding the problem of a globally uniform flight path being unable to adapt to the geometric differences of different parts of the blade. A flight path is designed specifically for each zone's geometric characteristics, improving flight path adaptability and inspection effectiveness. Specifically, the zone can be divided into a low-curvature smooth zone, a high-curvature edge zone, and an end-tip miniaturized zone. The low-curvature smooth zone corresponds to the blade root and main body of the blade; the high-curvature edge zone corresponds to the leading and trailing edges of the blade; and the end-tip miniaturized zone corresponds to the blade tip. The low-curvature smooth zone is characterized by a large cross-sectional dimension and a smooth surface at the blade root, which connects to the turbine hub. The end-tip miniaturized zone is characterized by an extremely small cross-sectional dimension and a drastic change in curvature at the blade tip. The high-curvature edge zone is characterized by the leading and trailing edges of the blade, which have a high curvature.

[0026] In one optional approach, multiple thresholds are preset, namely, a gentle curvature threshold, a high curvature threshold, and a miniaturized chord length threshold. The process iterates through all vertices of the three-dimensional mesh of the wind turbine blade. First, it determines whether the maximum chord length of the cross-section corresponding to each vertex is less than the miniaturized chord length threshold. If so, the region where the vertex is located is classified as the endpoint miniaturized region, corresponding to the tip of the blade. If not, it determines whether the absolute value of the maximum principal curvature of the vertex is greater than the high curvature threshold. If so, the region where the vertex is located is classified as the high curvature edge region, corresponding to the leading and trailing edges of the blade. If neither the chord length condition nor the high curvature condition is met, it determines whether the absolute value of the maximum principal curvature of the vertex is less than the gentle curvature threshold. If so, the region where the vertex is located is classified as the low curvature gentle region, corresponding to the root and main body of the blade. Through these determination rules, the entire wind turbine blade is divided into three different zones.

[0027] In another optional approach, based on the geometric characteristics of the wind turbine blade, the spanwise main axis of the blade is extracted, and the blade is divided into multiple segments along the spanwise main axis. For each segment, the average curvature and average chord length of all vertices within that segment are calculated. Based on the distribution patterns of the average curvature and average chord length, curvature intervals and chord length intervals are defined. Segments with average curvature in the low curvature interval and average chord length in the high chord length interval are classified as low curvature flat zones, corresponding to the blade root and blade body. Segments with average curvature in the high curvature interval are classified as high curvature edge zones, corresponding to the leading edge and trailing edge. Segments with average chord length in the minimum chord length interval are classified as tip miniaturization zones, corresponding to the blade tip. Finally, the wind turbine blade is divided into zones, ensuring that each zone corresponds to a specific part of the blade.

[0028] Step S30: Determine the route planning strategy suitable for each partition, and perform route planning for the corresponding partition based on the route planning strategy to obtain the route corresponding to each partition. The flight path planning strategy refers to a flight path generation method designed to adapt to the geometric features of different sections of the wind turbine blades. This strategy can include a set of specific planning rules such as setting waypoint step size, determining waypoint positions, and adjusting gimbal attitude. This strategy provides specific guidance for generating suitable flight paths for each section, ensuring that the flight path for each section conforms to the geometry of the area. This ensures both the quality of inspection imaging and avoids wasting battery life on redundant waypoints, achieving differentiated and refined flight path planning. For example, the flight path planning strategy for low-curvature, flat areas is a sparse grid strategy; for high-curvature edge areas, it is a high-density normal tracking strategy; and for miniaturized endpoint areas, it is a spherical multi-view envelope strategy. Each flight path planning strategy corresponds to different waypoint generation methods and gimbal control logic. For example, in the low curvature flat area, it consists of waypoints with large spacing, which are evenly distributed along the blade span, and the gimbal maintains an orthogonal shooting posture; in the high curvature edge area, the high-density flight path consists of waypoints with very small spacing, which are distributed along the blade edge ridge, and the gimbal adjusts its posture at high frequency to maintain orthogonality.

[0029] In one alternative approach, operators corresponding to different flight path planning strategies are preset for different partitions, with different execution logics for each operator. For example, for low-curvature, flat regions, a sparse grid operator based on overlap rate limits is selected as the appropriate flight path planning strategy; for high-curvature edge regions, a high-density normal tracking operator based on curvature inversion is selected as the appropriate flight path planning strategy; and for endpoint miniaturization regions, a multi-view envelope operator based on spherical coordinates is selected as the appropriate flight path planning strategy. By matching a dedicated flight path planning strategy to the geometric features of each partition, redundant waypoints in flat regions that waste UAV endurance are avoided, while the inspection accuracy and image overlap rate in high-curvature and leaf tip regions are ensured, blind spots are eliminated, and a high-quality segmented flight path foundation is provided for subsequent overall flight path stitching.

[0030] Step S40: Piece together the flight paths corresponding to each partition to obtain the target flight path corresponding to the UAV, where the target flight path is the flight path of the UAV when inspecting the wind turbine blades.

[0031] Splicing refers to the process of smoothly transitioning and integrating the independently generated flight paths of each section according to the actual structural sequence of the blades, eliminating gaps and attitude jumps between different section flight paths, and forming a continuous flight path. By integrating the segmented section flight paths into a complete flight path that can cover the entire wind turbine blade, it ensures that the UAV can smoothly transition from one section to another, avoiding problems such as flight instability or imaging discontinuities at the connection points of the flight paths, and ensuring the continuity of the inspection.

[0032] The target flight path refers to the complete flight path of the entire wind turbine blade, obtained by stitching together the flight paths of each section, covering the blade from root to tip. It is the flight path for the UAV to perform wind turbine blade inspection tasks, including all waypoints, attitude, speed, and other information. This target flight path guides the UAV to complete the entire wind turbine blade inspection operation, ensuring the UAV can complete the blade inspection and avoid blind spots and redundant flight.

[0033] In one optional approach, the connection sequence of the flight paths for each zone is first determined. Following the order of low-curvature gentle zones, high-curvature edge zones, and endpoint miniaturization zones, the starting and ending waypoints of adjacent zone flight paths are found. A B-spline smoothing algorithm is used to curve-fit the connecting waypoints of adjacent zones, generating smooth transition flight paths and eliminating abrupt changes in waypoint spacing and attitude jumps between adjacent flight paths. Then, all zone flight paths and transition flight paths are integrated into a continuous flight path. The waypoint sequence of the integrated flight path is verified to ensure that the waypoint sequence conforms to the actual structure of the blade and avoids waypoint confusion. Finally, the entire flight path is optimized by deleting redundant transition waypoints and adjusting the waypoint spacing to obtain the target flight path for UAV inspection. By integrating the independently generated segmented flight paths of each zone into a complete target flight path, the UAV can smoothly transition from one zone to another, achieving a comprehensive inspection of the entire wind turbine blade. This avoids problems such as flight instability or imaging tortuosity at flight path connections, while providing the UAV with a directly executable flight path, thus achieving the final goal of flight path planning.

[0034] In this embodiment, the geometric features of the wind turbine blades are first determined, and then the blades are divided into multiple partitions corresponding to different parts based on the geometric features. A suitable route planning strategy is matched for each partition and a route is generated independently. Finally, the routes of each partition are spliced ​​together to obtain the target route of the UAV. This scheme effectively solves the problem of uneven spatial sampling caused by a globally unified route, can adaptively match the geometric characteristics of the wind turbine blades, realize the differentiation of different regions and generate refined routes, and improve the accuracy of the generated routes.

[0035] In one feasible implementation, please refer to Figure 2 Step S10 may include steps S11 to S13: Step S11: Obtain the three-dimensional model of the wind turbine blade and determine the curvature of the vertices of each triangular facet. The three-dimensional model is composed of multiple triangular facets. Triangular facets are the basic units that make up the 3D model of a wind turbine blade. They are triangular planes composed of three vertices and three sides. Multiple triangular facets are pieced together according to certain topological relationships to form a complete 3D surface of the wind turbine blade, capable of accurately fitting the complex curved surface morphology of the blade. As a fundamental component of the 3D model, they carry the spatial coordinate information of the blade surface. The coordinate data of their vertices is the data source for calculating the curvature of the blade vertices, and is also the key carrier for finding the intersection of the slicing plane with the 3D model and extracting cross-sections. The 3D model of a wind turbine blade consists of hundreds or even thousands of triangular facets. Each triangular facet corresponds to a small area of ​​the blade surface. The size of the triangular facets can be adjusted according to the modeling accuracy. For areas of the blade surface with drastic curvature changes, such as the leading edge, trailing edge, and blade tip, the density of triangular facets is higher to ensure modeling accuracy. Each vertex of each triangular facet has corresponding spatial 3D coordinates, which are the objects for calculating the blade curvature and can reflect the spatial position information of a specific location on the blade surface.

[0036] Curvature is a quantitative parameter characterizing the degree and direction of curvature at each vertex of a wind turbine blade. It reflects the surface's unevenness and the severity of curvature, and is one of the parameters that distinguishes the geometric features of different parts of the blade. As an important component of blade geometry, it provides a basis for subsequent blade zoning. By analyzing the curvature differences at different vertices, it is possible to accurately distinguish between gentle, high-curvature, and miniature regions of the blade, while also providing data support for adapting flight path planning strategies. Specifically, curvature includes the maximum principal curvature, minimum principal curvature, and Gaussian curvature at the vertices. The vertices at the blade root and main body have smaller curvatures, indicating gentle surfaces in these areas; while the vertices at the leading edge, trailing edge, and blade tip have larger curvatures, indicating severe curvature in these areas.

[0037] In one alternative approach, a comprehensive scan of the wind turbine blade is performed using a laser 3D scanning device to obtain all spatial coordinate data of the blade surface. This data is then imported into 3D modeling software, where it is denoised, smoothed, and stitched to construct a 3D model of the wind turbine blade composed of multiple triangular facets. Next, each triangular facet in the 3D model is traversed, and the three vertices of each facet are extracted. For each vertex, a local quadratic surface fitting method is used to establish a local tangent plane coordinate system with the vertex as the origin. A K-DTree search is used to find the neighboring vertices of the vertex, and the local quadratic surface equation is fitted using the least squares method. The coefficients of the quadratic terms are extracted, and a Hessian matrix is ​​constructed. Eigenvalue decomposition is performed on the Hessian matrix to obtain the maximum principal curvature, minimum principal curvature, and Gaussian curvature of the vertex. This process is repeated for all triangular facet vertices to determine their curvature.

[0038] Step S12: Determine the chord length of each section of the wind turbine blades; A cross-section refers to the 3D cross-section of a wind turbine blade, obtained by cutting the blade's 3D model along its spanwise main axis using a plane perpendicular to the spanwise main axis. This cross-section reflects the shape and size of the blade at that cut location and is fundamental for calculating the chord length. The shape and size of the cross-section reflect the dimensional variation of the blade along its spanwise direction. The chord length is a crucial basis for distinguishing different parts of the blade, such as the root, body, and tip, and is also a component of the blade's geometric characteristics. Multiple cross-sections can be obtained by cutting along the spanwise main axis of the wind turbine blade from the root to the tip at fixed step lengths. The cross-section at the root is larger and relatively regular in shape, the cross-section at the body is of moderate size, and the cross-section at the tip shrinks sharply and becomes more pointed.

[0039] Chord length refers to the maximum width of a cross-section of a wind turbine blade, which is the straight-line distance between the two farthest vertices on the cross-sectional profile. It is a quantitative parameter characterizing the size of the blade cross-section and reflects the dimensional shrinkage pattern of the blade along its span. As an important component of the blade's geometry, chord length, together with vertex curvature, forms the basis for determining blade zoning. By observing changes in chord length, the miniature areas at the blade tip can be accurately identified. Specifically, the chord length is largest at the blade root, gradually decreases along the blade body, and is smallest at the blade tip.

[0040] In one optional approach, the spanwise principal axis of the 3D model of the wind turbine blade is first extracted. A series of slice planes perpendicular to the spanwise principal axis are generated along this axis at a preset fixed step size. Each slice plane is then intersected with the 3D model of the wind turbine blade, and the intersection points between each slice plane and the model's triangular facets are extracted. These intersection points are connected in topological order to form the cross-sectional profile of each slice position. A convex hull algorithm is then performed on each cross-sectional profile to obtain its convex hull profile. All vertex pairs on the convex hull profile are traversed, and the Euclidean distance between each pair of vertices is calculated. The distance between the two vertices with the largest distance is determined as the chord length of the cross-section. This process is repeated to determine the chord lengths of all cross-sections.

[0041] Step S13: Determine the curvature of the vertex and the chord length of the cross section as the geometric features of the wind turbine blade.

[0042] In one alternative approach, the curvature data of all triangular facet vertices are organized, and the maximum principal curvature, minimum principal curvature, and Gaussian curvature of each vertex are selected and stored according to the spatial position of the vertex. The chord length data of all cross sections are organized and sorted according to the spanwise order of the cross sections, and the position of each cross section and the corresponding chord length are recorded. The organized curvature data and chord length data are integrated to form a complete parameter set, which is used as the geometric feature of the wind turbine blade for subsequent partitioning and route planning.

[0043] In this embodiment, by acquiring a three-dimensional model composed of multiple triangular facets, the complex curved surface morphology and spatial structure of the wind turbine blade can be realistically reproduced, solving the problem of traditional geometric feature extraction lacking accurate model support. By determining the curvature of each triangular facet vertex, the differences in the degree of curvature of different parts of the blade surface can be accurately captured. Furthermore, by determining the chord length of each cross section, the dimensional variation law of the blade along the span can be clearly grasped, and the low-curvature gentle area, high-curvature edge area, and end-point miniature area can be accurately identified, providing a dimensional basis for subsequent partitioning.

[0044] Determining the curvature of the vertices of each triangular facet in step S11 above includes: Step S111: Traverse each vertex of each triangular facet. For each vertex, determine the normal vector of that vertex, and establish a local tangent plane coordinate system with that vertex as the origin and the normal vector of that vertex. The normal vector is a vector perpendicular to the tangent plane at a vertex on the surface of a wind turbine blade. It characterizes the spatial orientation of the blade surface at that vertex and serves as a parameter for establishing a local tangential plane coordinate system and determining the curvature direction of the surface. As the basis for establishing the local tangential plane coordinate system, it provides a spatial reference for the projection of neighboring vertices and the fitting of quadratic surfaces. Simultaneously, it helps characterize the curvature direction of the blade surface, ensuring the accuracy of curvature calculations; it is an intermediate parameter in the curvature calculation process. The normal vector direction at vertices in the low-curvature, gently curvature region is relatively stable with small variations. In contrast, the normal vector direction at vertices in the high-curvature edge region and blade tip changes drastically, reflecting the intense curvature characteristics of the surface in that region. The change in the normal vector direction is consistent with the curvature direction of the surface.

[0045] A local tangent plane coordinate system refers to a three-dimensional coordinate system with the target vertex as the origin, the vertex's normal vector as one of the coordinate axes, and two other mutually orthogonal coordinate axes constructed within the tangent plane of that vertex. This system is used for coordinate projection and quadratic surface fitting of local regions. By converting the global coordinates of the target vertex and its neighboring vertices into local coordinates, the three-dimensional surface is locally flattened, simplifying the quadratic surface fitting process, improving fitting accuracy, and facilitating subsequent construction of the Hessian matrix and eigenvalue decomposition, ensuring the accuracy of curvature calculation. Specifically, with the target vertex as the origin and the vertex's normal vector as the Z-axis, mutually orthogonal X-axis and Y-axis are constructed within the tangent plane perpendicular to the Z-axis, forming a local right-handed coordinate system. The global coordinates of neighboring vertices are projected onto the XY plane of this local tangent plane coordinate system, facilitating quadratic surface fitting.

[0046] In one alternative approach, all triangular facets in the 3D model of the wind turbine blade are traversed in the order from blade root to blade tip and from leading edge to trailing edge. The three vertices of each triangular facet are processed one by one. For each target vertex, the normal vector of the vertex is calculated using an area-weighted method. That is, all triangular facets adjacent to the vertex are selected, the normal vector of each adjacent triangular facet is calculated, and the normal vectors of all adjacent triangular facets are weighted according to their areas to obtain the normal vector of the target vertex. With the target vertex as the origin, the calculated normal vector is used as the Z-axis of the local tangent plane coordinate system. In the tangent plane perpendicular to the Z-axis, two mutually orthogonal unit vectors are constructed as the X-axis and Y-axis to form a local right-handed coordinate system, ensuring that the local tangent plane coordinate system can accurately fit the curved tangent plane at the target vertex.

[0047] Step S112: Search for the neighboring vertices of the vertex with the vertex as the center, and project all the neighboring vertices of the vertex onto the local tangent plane coordinate system; Neighborhood vertices refer to vertices adjacent to the target vertex. These vertices reflect the surface morphology around the target vertex and serve as auxiliary data for fitting a quadratic surface and calculating the curvature of the target vertex. A target vertex can have multiple neighboring vertices. Using the target vertex as the center, a search is performed with a preset radius to select the 20 vertices closest to it as neighborhood vertices. These neighborhood vertices are distributed around the target vertex, comprehensively reflecting the surface curvature trend at the target vertex and providing reliable data support for quadratic surface fitting.

[0048] Projection refers to the process of transforming the 3D coordinates of neighboring vertices into a local tangent plane coordinate system with the target vertex as the origin. This achieves local flattening of the 3D surface, eliminates the computational complexity caused by global coordinates, facilitates subsequent quadratic surface fitting, improves fitting efficiency and accuracy, and lays the foundation for curvature calculation. Specifically, the 3D coordinates of neighboring vertices are projected onto the XY plane (tangent plane) of the local tangent plane coordinate system, obtaining the 2D coordinates of each neighboring vertex in the XY plane. These 2D coordinates reflect the planar distribution of neighboring vertices relative to the target vertex, facilitating quadratic surface fitting.

[0049] In one optional approach, neighboring vertices are searched around the target vertex at a preset search radius. The search radius is set based on the accuracy and noise level of the blade's 3D model to ensure that the searched neighboring vertices fully reflect the surface morphology around the target vertex while avoiding interference from distant vertices in the fitting results. After the search is complete, the 3D coordinates of all neighboring vertices are obtained. Using a coordinate transformation algorithm, the 3D coordinates of each neighboring vertex are projected onto a local tangent plane coordinate system with the target vertex as the origin, converting the 3D coordinates into 2D or 3D coordinates in the local tangent plane coordinate system. Crucially, the projected coordinates of the neighboring vertices within the tangent plane (XY plane) of the local tangent plane coordinate system are preserved, providing data support for subsequent quadratic surface fitting. By acquiring the surface morphology data around the target vertex and converting the 3D data into simplified data in the local tangent plane coordinate system through projection, it is easier to fit the subsequent quadratic surface. Simultaneously, K-DTree search can quickly and accurately obtain neighboring vertices, improving computational efficiency, avoiding interference from redundant data, and ensuring the accuracy of the fitted data.

[0050] Step S113: In the local tangent plane coordinate system, the quadratic surface equation is fitted by the least squares method based on the projected coordinates of all neighboring vertices of the vertex, and the coefficients of the quadratic term of the quadratic surface equation are extracted. Least squares is a mathematical method for data fitting. By minimizing the sum of squared errors between the fitted values ​​and the actual observed values, the coefficients of the fitting equation are determined, thus obtaining a fitted curve or surface that best approximates the actual data distribution. When used to fit quadratic surface equations based on the projected coordinates of neighboring vertices, it effectively reduces the impact of local scanning noise on the fitting results, ensuring that the fitted quadratic surface accurately reflects the surface morphology at the target vertex. This provides accurate equation coefficients for subsequent construction of the Hessian matrix and curvature calculation.

[0051] The quadratic surface equation is a mathematical equation used to describe the surface morphology at a target vertex. It includes quadratic terms, linear terms, and constant terms. The coefficients of the quadratic terms reflect the degree of curvature of the surface and are the basis for constructing the Hessian matrix and calculating curvature. It is used to quantitatively characterize the surface morphology at the target vertex, and its quadratic coefficients are the data for subsequently constructing the Hessian matrix. By fitting the quadratic surface equation, the surface morphology around the target vertex can be transformed into a mathematical expression, facilitating the calculation and analysis of curvature.

[0052] The quadratic coefficients refer to the coefficients in the equation of a quadratic surface that include the squared terms and product terms of the variables. Specifically, they include the coefficients of the x-squared term, the x-multiplied by y term, and the y-squared term. These coefficients directly determine the degree and direction of curvature of the quadratic surface. The values ​​of the quadratic coefficients directly reflect the degree of curvature of the surface at the target vertex. By constructing the Hessian matrix through the quadratic coefficients, the curvature can be solved, serving as an intermediate parameter connecting quadratic surface fitting and curvature calculation.

[0053] In one optional approach, under the local tangent plane coordinate system, the projected coordinates of all neighboring vertices are substituted into a predefined quadratic surface equation. This equation includes x-squared terms, x-by-y terms, y-squared terms, x terms, y terms, and a constant term. Based on the substituted projected coordinates, an overdetermined linear equation system is constructed, where the unknowns are the coefficients of each term in the quadratic surface equation. The overdetermined linear equation system is solved using the least squares method. By minimizing the sum of squared errors between the fitted values ​​and the actual projected coordinates of the neighboring vertices, all coefficients of the quadratic surface equation are determined. After solving, the quadratic coefficients—namely, the x-squared, x-by-y, and y-squared coefficients—are extracted from the equation for subsequent Hessian matrix construction. By transforming the surface morphology around the target vertex into a mathematical expression, the least squares fitting method effectively eliminates the influence of local scanning noise, ensuring that the quadratic surface equation accurately reflects the surface curvature characteristics at the target vertex. The extracted quadratic coefficients provide reliable support for subsequent Hessian matrix construction and curvature calculation.

[0054] Step S114: Construct the Hessian matrix based on the coefficients of the quadratic terms, and perform eigenvalue decomposition on the Hessian matrix to obtain the curvature of the vertex, where the Hessian matrix characterizes the degree of curvature of the surface at the vertex.

[0055] The Hessian matrix is ​​a second-order symmetric matrix constructed from the quadratic coefficients of the equation of a quadratic surface. It is a mathematical tool characterizing the second-order differential properties of a surface at a target vertex, quantifying the degree and direction of curvature. As a vehicle for solving the curvature of a target vertex, by performing eigenvalue decomposition on the Hessian matrix, eigenvalues ​​characterizing the degree of surface curvature can be obtained, thus determining the curvature of the target vertex. It is a mathematical tool in the curvature calculation process. For example, there exists a Hessian matrix constructed from the quadratic coefficients a, b, and c, where the first row and first column is 2a, the first row and second column is b, the second row and first column is b, and the second row and second column is 2c. The symmetric structure of this Hessian matrix accurately reflects the curvature characteristics of the surface in two perpendicular directions.

[0056] In one optional approach, a Hessian matrix is ​​constructed based on the extracted quadratic coefficients. The first row and first column of the Hessian matrix are twice the coefficient of the x-squared term in the quadratic equation; the first row and second column are the coefficients of the x-y term; the second row and first column are the coefficients of the x-y term; and the second row and second column are twice the coefficients of the y-squared term. After construction, the Hessian matrix is ​​decomposed using an eigenvalue decomposition algorithm to obtain two eigenvalues. The eigenvalue with the larger value is determined as the maximum principal curvature of the target vertex, and the eigenvalue with the smaller value is determined as the minimum principal curvature of the target vertex. Simultaneously, the product of the two eigenvalues ​​is calculated to obtain the Gaussian curvature of the vertex, thus completing the determination of the vertex curvature. By transforming the quadratic coefficients of the quadratic surface equation into curvature parameters characterizing the degree of surface bending, eigenvalue decomposition can accurately obtain the maximum, minimum, and Gaussian curvatures of the target vertex, realizing the quantification of the surface bending characteristics. This provides data for subsequent determination of blade geometric features and partitioning, ensuring that the geometric features accurately reflect the surface morphology of the blade.

[0057] In this embodiment, by traversing all vertices of the triangular facets, the comprehensiveness of curvature calculation is ensured, avoiding inaccurate geometric features caused by missing vertices and providing complete data support for the subsequent determination of geometric features. By determining the vertices and normal vectors and establishing a local tangent plane coordinate system, local flattening of the 3D surface is achieved, simplifying subsequent data processing and fitting processes. Simultaneously, it ensures that the fitting results accurately reflect the surface morphology at the target vertex, improving the accuracy of curvature calculation. Using K-DTree search to find neighboring vertices and projecting them allows for rapid and accurate acquisition of surface data around the target vertex, effectively avoiding interference from distant vertices and noise points, providing a reliable data foundation for quadratic surface fitting. Fitting the quadratic surface equation using the least squares method effectively eliminates the influence of local scanning noise, ensuring that the fitted surface truly reflects the bending characteristics at the target vertex. The extracted quadratic coefficients provide accurate intermediate parameters for subsequent curvature calculations. Finally, by constructing the Hessian matrix and performing eigenvalue decomposition, the maximum principal curvature, minimum principal curvature, and Gaussian curvature of the vertices were accurately obtained, realizing the quantification of the bending characteristics of the surface. This provides a basis for the bending characteristics of subsequent blade partitioning and flight path planning, solving the problems of low accuracy and poor noise resistance of traditional curvature calculation methods. It improves the accuracy and reliability of the entire flight path planning method, ensuring that subsequent partitioning can accurately distinguish different regions of the blade and that flight path planning can conform to the surface shape of the blade.

[0058] Determining the chord length of each section of the wind turbine blades in step S12 above includes: Step S121: Determine the spanwise main axis of the wind turbine blade, and generate multiple slice planes orthogonal to the spanwise main axis along the spanwise main axis according to a preset step size; The spanwise principal axis refers to the axis running through the root to the tip of the wind turbine blade. It characterizes the length direction of the blade and serves as the baseline for the blade's spanwise direction. Its direction is consistent with the blade's length direction, reflecting the overall spatial attitude of the blade. As the reference for generating the slice plane, it determines the generation direction and orthogonality of the slice plane, ensuring that the slice plane can be uniformly distributed along the blade's spanwise direction. This allows for accurate acquisition of the cross-section at various positions along the blade's spanwise direction, providing a reference for calculating the chord length of the cross-section.

[0059] The preset step size refers to the distance between two adjacent slice planes when generating slice planes along the spanwise axis of the wind turbine blade. This distance can be preset according to the blade's geometric characteristics and the required computational accuracy. This preset step size controls the generation density of the slice planes. A smaller step size results in denser slice planes and more detailed cross-sectional information; a larger step size results in sparser slice planes and higher computational efficiency. The preset step size balances computational accuracy and efficiency, ensuring the accuracy and efficiency of cross-sectional chord length calculation. For example, based on the blade's geometric characteristics, in the low-curvature, gently curvature region, the preset step size can be set to 50 cm to ensure sufficient cross-sectional information is obtained; in the tip miniaturization region, the preset step size can be set to 20 cm to increase cross-sectional density and accurately capture changes in blade tip dimensions.

[0060] A slicing plane refers to a plane perpendicular to the spanwise main axis of a wind turbine blade, generated at a preset step size. It serves as a tool for cutting the 3D model of the wind turbine blade and obtaining its cross-section, accurately reflecting the cross-sectional shape of the blade at a specific location. As the carrier for obtaining the blade's cross-section, by intersecting with the 3D model of the blade, it can obtain the cross-sectional contours of the blade at various locations along the spanwise direction, providing a basis for calculating the chord length and ensuring a comprehensive understanding of the blade's dimensional changes along the spanwise direction. Specifically, multiple slicing planes are generated along the spanwise main axis at preset step sizes. Each slicing plane is perpendicular to the spanwise main axis and distributed sequentially from the blade root to the blade tip. After each slicing plane cuts the 3D model of the wind turbine blade, a corresponding cross-section is obtained. These cross-sections collectively reflect the dimensional and shape changes of the blade along the spanwise direction.

[0061] In one optional approach, the root center point and tip center point of the 3D model of the wind turbine blade are first extracted, and the two center points are connected to obtain the preliminary spanwise principal axis. The preliminary spanwise principal axis is then optimized and adjusted to ensure that the principal axis can penetrate the area of ​​the blade, is consistent with the length direction of the blade, and conforms to the overall spatial attitude of the blade. According to the geometric characteristics of the blade and the calculation accuracy requirements, an appropriate slicing step size is preset. The step size can be appropriately increased in the low curvature gentle area and appropriately decreased in the end point miniaturized area. Along the optimized spanwise principal axis, from the blade root to the blade tip, multiple slicing planes are generated sequentially according to the preset step size. Each slicing plane is orthogonal to the spanwise principal axis to ensure that the slicing plane can cut the blade perpendicularly and obtain accurate cross-sectional information.

[0062] Step S122: Intersect each slice plane with the three-dimensional model of the wind turbine blade to obtain the cross section at each slice position. The cross section is composed of the set of intersection points between the corresponding slice plane and the three-dimensional model. Intersection calculation refers to determining the intersection between the slice plane and the 3D model of the wind turbine blade. Specifically, it involves identifying all intersection points between the slice plane and the triangular facets of the 3D model of the wind turbine blade as the slice plane passes through it. These intersection points collectively form the profile of the cross-section. By obtaining the profile point set of the wind turbine blade cross-section, raw data is provided for subsequent extraction of the cross-section profile and calculation of the chord length, ensuring that the cross-section accurately reflects the cross-sectional shape of the blade at that location. This is achieved by performing intersection calculations on each slice plane and the 3D model of the wind turbine blade, calculating the intersection between the slice plane and each triangular facet in the 3D model. If the slice plane intersects with a triangular facet, the coordinates of the intersection point are obtained. All intersection points are collected to form the cross-section point set for that slice location.

[0063] The intersection point set refers to the set of all intersection points obtained after intersecting the slice plane with the 3D model of the wind turbine blade. These points are arranged in a certain topological order, forming a complete cross-sectional profile that reflects the cross-sectional shape of the blade at that slice location. As the raw data of the cross-sectional profile, it is the basis for subsequent convex hull processing and chord length calculation. By processing the intersection point set, a clear and complete cross-sectional profile can be obtained, ensuring the accuracy of the chord length calculation.

[0064] In one alternative approach, each generated slice plane is intersected with the 3D model of the wind turbine blade. All triangular faces in the 3D model of the blade are traversed, and it is determined whether each triangular face intersects with the current slice plane. If a triangular face intersects with the slice plane, the 3D coordinates of the intersection point are calculated, and all intersection points are collected to form the intersection point set for that slice location. The intersection point set is then denoised to remove outliers caused by model errors or calculation errors. Finally, according to the topological relationship of the triangular faces, the corner points in the intersection point set are arranged in order to form a complete cross section, ensuring that the cross section can accurately reflect the cross-sectional shape of the blade at that slice location.

[0065] Step S123: Perform convex hull processing on each cross section to obtain the convex hull contour of each cross section; Convex hull processing refers to processing the set of intersection points of a cross section to extract the convex hull profile. This involves finding the smallest convex polygon that can enclose all intersection points, eliminating pits, jagged edges, and noise points in the point set, and preserving the outer contour of the cross section. It is used to eliminate noise and outliers in the intersection point set of a cross section, simplify the cross section profile, preserve the outer contour characteristics of the cross section, and ensure that subsequent chord length calculations accurately reflect the maximum width of the cross section, avoiding chord length calculation errors caused by noise points.

[0066] The convex hull profile refers to the smallest convex polygon obtained after convex hull processing that can enclose all intersection points of the cross section. It is a simplified form of the cross section's outer contour, clearly reflecting the maximum width and overall shape of the cross section, and eliminating internal noise and pits. It is used for subsequent chord length calculation. By determining the vertex distances on the convex hull profile, the maximum chord length of the cross section can be accurately obtained, ensuring the accuracy and reliability of the chord length calculation.

[0067] In one alternative approach, a convex hull algorithm is performed on the intersection point set of each cross section. First, the coordinates of the intersection point set are standardized to ensure calculation accuracy. Then, the convex hull algorithm is used to find the smallest convex polygon that can enclose all intersection points of the cross section. The vertices of this convex polygon are the boundary points in the intersection point set. Points and noise points located inside the convex polygon in the intersection point set are removed, and the vertices of the convex polygon are retained. These vertices are connected in sequence to form the convex hull contour of the cross section. The convex hull contour is smoothed to eliminate sharp transitions between vertices, ensuring that the convex hull contour can clearly and accurately reflect the outer contour shape of the cross section.

[0068] Step S124: For each convex hull profile, determine the distance between every two vertices on the convex hull profile, and determine the chord length of the cross section corresponding to the convex hull profile based on the two vertices with the largest distance.

[0069] In one alternative approach, for the convex hull profile of each cross section, all vertices on the profile are traversed, and every two different vertices are selected in turn. The Euclidean distance calculation formula is used to calculate the straight-line distance between the two vertices. During the calculation process, the distance values ​​between all vertex pairs are recorded in real time, and the two vertices with the largest distance are selected. The distance between these two vertices is determined as the chord length of the cross section.

[0070] In this embodiment, by determining the main spanwise axis of the blade and generating orthogonal slicing planes according to a preset step size, it is ensured that the slicing planes are uniformly distributed along the spanwise direction of the blade, fully covering all parts of the blade, including the root, blade body, and tip. This provides a reasonable benchmark for subsequent cross-section acquisition and avoids missing or redundant chord length data caused by uneven distribution of slicing planes. Secondly, by finding the intersection of the slicing planes with the 3D model, the intersection point set of the cross-section at each slice position can be accurately obtained. Denoising and sorting processes ensure the accuracy and completeness of the intersection point set, providing reliable raw data for subsequent convex hull processing and chord length calculation, solving the problems of low accuracy and high noise in traditional cross-section extraction. Thirdly, convex hull processing eliminates noise points and pits in the cross-sectional contour, preserving the outer contour features of the cross-section, simplifying the chord length calculation process, and avoiding chord length calculation errors caused by noise points, thus ensuring the accuracy of chord length calculation. Finally, by traversing the vertices of the convex hull profile to calculate the maximum distance, the maximum chord length of each section can be accurately obtained, and the dimensional variation law of the blade along the span can be clearly understood. This provides an important dimensional basis for the subsequent determination of blade geometry and partitioning, accurately identifies the endpoint miniature area, provides a dimensional reference for the adaptation of flight path planning strategies, improves the accuracy and reliability of the entire flight path planning method, ensures that the subsequent partitioning can accurately distinguish different areas of the blade, and ensures that the flight path planning can conform to the dimensional variation law of the blade, avoiding problems such as unreasonable partitioning and poor flight path adaptability caused by inaccurate chord length calculation.

[0071] In one feasible implementation, step S20 includes step S21: Step S21: Based on the conditions satisfied by the geometric features, the wind turbine blades are divided into different zones; The partitions include a first partition, a second partition, and a third partition. The geometric features of the first partition satisfy the following conditions: the absolute value of the curvature of the vertex is less than a first set value, and the chord length of the cross section is greater than a preset chord length. The geometric features of the second partition satisfy the following conditions: the absolute value of the curvature of the vertex is greater than a second set value. The geometric features of the third partition satisfy the following conditions: the chord length of the cross section shows a decreasing trend, and the chord length of the cross section is less than a preset chord length. The first set value is less than the second set value.

[0072] The first zone, also known as the low curvature and gentle area, refers to the region in the wind turbine blade where the absolute value of the vertex curvature is less than the first set value and the chord length of the cross section is greater than the preset chord length. The geometric characteristics of this region are a gentle curvature and a large cross-sectional size, which is the main area of ​​the blade.

[0073] The second zone, namely the high curvature edge zone mentioned above, refers to the area in the wind turbine blade where the absolute value of the vertex curvature is greater than the second set value. The geometric characteristics of this area are characterized by severe surface bending and large curvature, and it is mostly located at the edge of the blade.

[0074] The third zone, also known as the aforementioned end-point miniaturization zone, refers to the area in the wind turbine blade where the chord length of the cross section decreases and is less than the preset chord length. The geometric characteristics of this area are characterized by a sharp reduction in cross-sectional size and a contracted state, mainly at the blade tip.

[0075] The first set value refers to the pre-set curvature threshold used to determine the first zone. It is the critical value that distinguishes the low curvature flat zone from other areas. Its value is determined according to the design parameters of the wind turbine blades, the modeling accuracy, and the inspection requirements, and is less than the second set value.

[0076] The second set value refers to the pre-set curvature threshold used to determine the second zone. It is the critical value that distinguishes the high curvature area of ​​the blade from other areas. Its value is determined according to the design parameters of the wind turbine blade, the modeling accuracy, and the inspection requirements, and is greater than the first set value.

[0077] The preset chord length refers to the chord length threshold set in advance to determine the first and third zones. It is the critical value that distinguishes the main body area of ​​the blade from the miniature end area. Its value is determined according to the design parameters of the wind turbine blade, the distribution law of cross-sectional size, and the inspection requirements.

[0078] In one optional approach, the geometric features of the wind turbine blade are first organized, clarifying the curvature of all vertices and the chord length of all cross sections. Simultaneously, preset first and second set values ​​and preset chord lengths are retrieved. Then, all vertices and cross sections of the blade's 3D model are traversed. For each vertex's region, it is first determined whether the absolute value of the curvature of the vertex in that region is less than the first set value, and simultaneously whether the chord length of the corresponding cross section in that region is greater than the preset chord length. If so, the region is divided into the first partition. If the first partition condition is not met, it is determined whether the absolute value of the curvature of the vertex in that region is greater than the second set value. If so, the region is divided into the second partition. If neither the first nor the second partition condition is met, it is determined whether the chord length of the cross section in that region shows a decreasing trend and is less than the preset chord length. If so, the region is divided into the third partition. After traversal, the boundaries of the three partitions are smoothed to ensure natural transitions between partitions, with no overlap or omissions, thus completing the blade partitioning. By precisely dividing the wind turbine blades into three independent zones based on their geometric characteristics, the geometric features and spatial range of each zone are clearly defined. This lays the foundation for adapting each zone to a dedicated flight path planning strategy, avoiding the problem that a globally unified flight path cannot adapt to the geometric differences of different parts of the blade. This ensures the targeted and refined nature of the flight path planning, while also ensuring the smooth implementation of the entire flight path planning method.

[0079] In this embodiment, the geometric features are explicitly defined as vertex curvature and cross-sectional chord length, making the basis for partitioning more specific and clear. Secondly, the blade is divided into three partitions, each corresponding to a different part of the blade, and the geometric feature judgment conditions for each partition are clearly defined. This allows for precise differentiation of the low-curvature flat area, high-curvature region, and tip miniature area of ​​the blade, solving the problem of inaccurate partitioning in traditional methods. Thirdly, by setting a first setpoint, a second setpoint, and a preset chord length, combined with the absolute value of curvature and the trend of chord length changes, a clear and quantifiable partitioning judgment logic is formed, ensuring the consistency and accuracy of partitioning. Finally, the completion of partitioning provides a clear regional basis for subsequent adaptation of each partition to the corresponding flight path planning strategy. This allows subsequent flight path planning to be differentiated based on the geometric characteristics of each partition, ensuring both the inspection accuracy of the high-curvature edge area and the tip miniature area, while also considering the inspection efficiency of the low-curvature flat area. This effectively improves the accuracy of the entire flight path planning method and provides support for the refined design of UAV inspection routes.

[0080] When the partition is the first partition, the above step S30 includes steps A31 to A34: Step A31: Determine the effective coverage width of a single image on the blade surface based on the preset safe flight distance, the field of view of the camera on the UAV, and the minimum image overlap rate, and determine the first waypoint step size based on the effective coverage width and the minimum image overlap rate. The preset safe flight distance refers to the minimum safe distance between the drone and the surface of the wind turbine blades during flight. This preset safe flight distance can be determined based on the size of the drone, flight stability, blade surface undulations, and inspection accuracy requirements to ensure that the drone does not collide with the blades during flight.

[0081] The camera field of view is the spatial range angle that the camera on the drone can capture. It is divided into horizontal field of view and vertical field of view. It is a parameter for calculating the effective coverage width of the image and determines the range of the blade surface that a single image can cover.

[0082] The minimum image overlap rate ensures that the inspected images can be completely stitched together without blind spots. It is a preset minimum overlap ratio between two adjacent inspected images, and its value is determined according to the inspection accuracy requirements. It is an important parameter for calculating waypoint step length. For example, based on the inspection accuracy requirements, the minimum image overlap rate can be preset to 80%, that is, the overlapping part of two adjacent images is not less than 80% of the effective coverage width of a single image. By constraining the waypoint step length with this parameter, the integrity of image stitching and the comprehensiveness of inspection are ensured.

[0083] The effective coverage width of a single image on the blade surface refers to the actual width of the wind turbine blade surface that a single image captured by the camera at a preset safe flight distance can cover. It is calculated from the preset safe flight distance and the camera's field of view, and serves as the basis for determining the waypoint step size. The larger the effective coverage width, the larger the waypoint step size can be set, resulting in higher inspection efficiency; conversely, the smaller the effective coverage width, the smaller the waypoint step size needs to be set to ensure that the image overlap rate meets requirements, thus providing a basis for determining the waypoint step size.

[0084] The first waypoint step size, designed for the first zone, is the distance between two adjacent waypoints. It is calculated from the effective coverage width of a single image and the minimum image overlap rate, and serves as a parameter for waypoint distribution within the first zone's flight path, determining the density of waypoints. For example, if the effective coverage width of a single image is 0.346 meters and the minimum image overlap rate is 80%, the calculated first waypoint step size is 0.069 meters, meaning the distance between two adjacent waypoints is 0.069 meters, ensuring that image overlap meets requirements and waypoint density is reasonable.

[0085] In one optional approach, the preset safe flight distance, the UAV camera's field of view parameters, and the minimum image overlap rate are first retrieved. The camera's field of view is a horizontal field of view, and the minimum image overlap rate is set according to the inspection accuracy requirements of the first zone. The effective coverage width of a single image on the blade surface is calculated using trigonometric functions. The preset safe flight distance is considered one leg of a right triangle, and half of the camera's field of view is an acute angle. Half of the effective coverage width is calculated using the tangent function, and then multiplied by 2 to obtain the complete effective coverage width. Then, based on the effective coverage width and the minimum image overlap rate, the first waypoint step size is calculated. The formula is: the first waypoint step size equals the effective coverage width multiplied by (1 minus the minimum image overlap rate). This ensures that images captured by adjacent waypoints meet the minimum overlap requirement while avoiding overly dense waypoints. After calculation, the first waypoint step size is verified to ensure it meets the UAV's flight control accuracy. If it does not, the minimum image overlap rate is adjusted, and the calculation is repeated until the requirements are met.

[0086] Step A32: Unfold the three-dimensional curved surface of the wind turbine blade into a two-dimensional UV plane along the spanwise main axis of the wind turbine blade. Generate a uniform grid in the two-dimensional UV plane according to the first waypoint step size, and project each grid in reverse onto the three-dimensional curved surface to obtain each surface anchor point. The three-dimensional curved surface of a wind turbine blade refers to the surface of the three-dimensional model of the wind turbine blade. The curved surface, formed by piecing together multiple triangular facets, can realistically reflect the gentle curved surface shape of the first zone and serves as the basic carrier for waypoint planning. As a spatial reference for waypoint generation, its gentle curved surface characteristics facilitate two-dimensional unfolding and grid projection, ensuring that the generated waypoints can fit the blade surface and providing support for the generation of routes in the first zone.

[0087] A two-dimensional UV plane refers to the two-dimensional plane obtained by unfolding the three-dimensional surface of the first section of a wind turbine blade along its spanwise principal axis. The U-axis corresponds to the spanwise direction of the blade, and the V-axis corresponds to the chordwise direction. This plane converts the spatial coordinates of the three-dimensional surface into two-dimensional coordinates, simplifying grid generation and waypoint planning. By converting the complex three-dimensional surface into a simple two-dimensional UV plane, it is easier to generate a uniform grid according to the first waypoint step size, reducing the complexity of waypoint planning. Simultaneously, it ensures that the grid can uniformly cover the entire surface of the first section, providing a basis for anchor point projection. For example, the three-dimensional surface of the first section can be unfolded along its spanwise principal axis to obtain a rectangular two-dimensional UV plane, with the U-axis extending along the spanwise direction and the V-axis extending along the chordwise direction. This plane can completely represent the surface range of the first section, facilitating the generation of a uniform grid according to the step size.

[0088] A grid refers to a uniform mesh generated in a 2D UV plane according to the first waypoint step size. It consists of a series of mutually perpendicular lines, and the intersections of the grid are the basis for subsequent projection to generate surface anchor points, ensuring that the anchor points are evenly distributed. As the basis for surface anchor point generation, it ensures that the anchor points are evenly distributed in the 2D UV plane according to the first waypoint step size, thereby ensuring that the subsequently generated waypoints are evenly distributed on the 3D curved surface of the first partition, avoiding blind spots or redundancy caused by uneven waypoint distribution.

[0089] Surface anchor points are spatial coordinate points obtained by inversely projecting the intersection points of a grid in a two-dimensional UV plane onto the three-dimensional curved surface of the first partition. Located on the blade surface, they are the basis for generating waypoints and reflect the projected position of the waypoints on the blade surface. As the reference point for waypoint generation, their spatial coordinates accurately reflect the corresponding position of the waypoints on the blade surface. By offsetting them along the normal vector by a preset safety distance, waypoints that meet safety requirements can be generated, providing support for waypoint generation. Each surface anchor point is located on the blade surface of the first partition, evenly distributed on a gentle curved surface, and its coordinates correspond one-to-one with the intersection points of the two-dimensional grid. After offsetting, they become waypoints for the UAV's flight.

[0090] In one optional approach, the spanwise principal axis of the wind turbine blade is first extracted, ensuring that the principal axis runs through the area of ​​the first partition and is consistent with the blade spanwise direction. The three-dimensional surface of the first partition is unfolded along the spanwise principal axis, and the spatial coordinates of the three-dimensional surface are converted into two-dimensional UV plane coordinates, where the U-axis corresponds to the blade spanwise direction and the V-axis corresponds to the blade chordwise direction, ensuring that the unfolded two-dimensional plane can completely and accurately reflect the surface range of the first partition. Within the two-dimensional UV plane, mutually perpendicular uniform grids are generated according to the determined first waypoint step size, and the intersections of the grids are potential anchor points, ensuring that the grids can uniformly cover the entire two-dimensional UV plane. Then, according to the unfolded correspondence between the three-dimensional surface and the two-dimensional UV plane, each grid intersection is back-mapped onto the three-dimensional surface of the first partition, and the three-dimensional coordinates corresponding to each intersection are obtained. These three-dimensional coordinates are determined as surface anchor points, ensuring that each surface anchor point accurately fits the three-dimensional surface of the first partition.

[0091] Step A33: Offset each surface anchor point outwards along the normal vector of its corresponding vertex by a preset safe flight distance to generate waypoints; In one optional approach, the normal vector of the blade vertex corresponding to the location of each surface anchor point is first obtained, ensuring that the normal vector is a unit vector perpendicular to the blade surface and pointing outwards. Then, the 3D coordinates of each surface anchor point are multiplied by the normal vector to obtain an offset vector, the length of which is equal to a preset safe flight distance. The 3D coordinates of each surface anchor point are added to the corresponding offset vector to obtain new 3D coordinates, which are then used as waypoints. After generating all waypoints, each waypoint is verified to determine if the distance between the waypoint and the blade surface equals the preset safe flight distance. If not, the direction of the normal vector or the offset distance is adjusted to ensure that all waypoints meet the safe flight requirements. By converting surface anchor points into UAV waypoints that meet safety requirements, the preset safe distance between the waypoints and the blade surface is maintained, preventing collisions between the UAV and the blade during flight. Simultaneously, the waypoint positions ensure that the camera can clearly capture images of the blade surface, providing qualified waypoint data for subsequent flight path generation.

[0092] Step A34: Generate the route corresponding to the first partition based on the waypoints corresponding to each surface anchor point.

[0093] In one alternative approach, all generated waypoints are first sorted according to the blade spanwise order from the blade root to the blade tip and the chordwise order from one side to the other. This ensures that the waypoint order conforms to the UAV's flight logic and avoids flight instability caused by waypoint misalignment. Then, curve fitting is performed on the sorted waypoints to eliminate angle jumps and distance abrupt changes between adjacent waypoints, making the transition between waypoints smoother. Finally, the smoothed waypoints are connected in sequence to form a complete flight path corresponding to the first partition. The integrity of the flight path corresponding to the first partition is verified to ensure that the flight path can cover the entire surface of the first partition without blind spots, while also ensuring that the flight path conforms to the flight characteristics of the UAV.

[0094] In this embodiment, by presetting a safe flight distance, camera field of view, and minimum image overlap rate, the effective coverage width and first waypoint step size are scientifically calculated. This ensures both the overlap rate of adjacent images, avoiding blind spots in the inspection, and prevents the waste of UAV endurance due to overly dense waypoints, achieving a balance between inspection accuracy and efficiency. Secondly, the three-dimensional curved surface is unfolded into a two-dimensional UV plane along the spanwise principal axis, simplifying the complexity of grid generation and waypoint planning. Simultaneously, inverse projection ensures that surface anchor points accurately fit the blade surface, providing a reliable benchmark for waypoint generation and ensuring uniform and accurate waypoint distribution. Thirdly, waypoints are generated by offsetting the surface anchor points along the normal vector by a preset safe distance, ensuring that the UAV maintains a safe distance from the blade during flight, avoiding collision risks, while also ensuring a reasonable camera shooting angle to clearly capture blade surface details, meeting the inspection accuracy requirements. Finally, by sorting, smoothing, and connecting waypoints, a complete route for the first zone is generated, ensuring that the UAV can complete the inspection operation of the first zone evenly and smoothly, adapting to the flat geometric characteristics of the first zone, solving the problem of redundancy or uneven coverage of waypoints in flat areas in traditional unified routes, improving the inspection efficiency and quality of the first zone, and providing important support for the refined planning of the entire wind turbine blade inspection route.

[0095] When the partition is the second partition, the curvature includes the maximum principal curvature, and each vertex has a corresponding maximum principal curvature; the above step S30 includes steps B31~B34: Step B31: Determine the second waypoint step size based on the maximum permissible line-of-sight deflection angle and radius of curvature, wherein the radius of curvature is positively correlated with the second waypoint step size; The maximum principal curvature refers to the curvature in the direction of the greatest curvature at a certain vertex on the blade surface. It is a parameter of the curvature at that vertex. The larger the value, the more severe the curvature at that vertex. It is the basis for determining the second zone and the waypoint step size.

[0096] The maximum permissible line-of-sight deflection angle refers to the maximum permissible angle between the pre-set line-of-sight of the UAV camera and the normal vector of the blade surface. Its value is determined based on the inspection accuracy requirements and camera performance to ensure that the camera can clearly capture the details of the blade surface. It is an important constraint for determining the second waypoint step size.

[0097] The radius of curvature refers to the radius corresponding to the maximum principal curvature at the apex. It is an auxiliary parameter that characterizes the degree of curvature of the blade surface. Numerically, it is equal to the reciprocal of the maximum principal curvature. The larger the maximum principal curvature, the smaller the radius of curvature, and the more severe the curvature.

[0098] The second waypoint step size refers to the distance between two adjacent waypoints designed for the second zone. It is calculated from the maximum permissible line-of-sight deflection angle and the radius of curvature, and is positively correlated with the radius of curvature. It is a parameter for the distribution of waypoints in the second zone's routes.

[0099] In one optional approach, firstly, the preset maximum permissible line-of-sight deflection angle and the maximum principal curvature of each vertex within the second partition are retrieved. The radius of curvature corresponding to each vertex is calculated using the maximum principal curvature, and the radius of curvature is equal to the reciprocal of the maximum principal curvature. Then, a second waypoint step length calculation model is established. The maximum permissible line-of-sight deflection angle and the radius of curvature are substituted into this model, and the second waypoint step length is calculated using trigonometric functions. Specifically, the calculation process involves using the radius of curvature as the hypotenuse and half of the maximum permissible line-of-sight deflection angle as the acute angle, and calculating the chord length of adjacent waypoints on the curvature arc using a sine function. The chord length is the second waypoint step size. Since the radius of curvature is positively correlated with the second waypoint step size, it's crucial to ensure that during calculation, the second waypoint step size increases synchronously when the radius of curvature increases and decreases synchronously when the radius of curvature decreases. After calculation, the second waypoint step sizes for all vertices are consolidated, and the average value within the same region is taken as the final second waypoint step size for that region. This ensures the stability of the waypoint step size and verifies whether it meets the UAV flight control accuracy. If not, the maximum allowable line-of-sight deflection angle is adjusted, and the calculation is recalculated until the requirements are met. The second waypoint step size is designed to adapt to the high curvature characteristics of the second region; a smaller radius of curvature results in a smaller second waypoint step size and a higher waypoint density, ensuring that the camera's line-of-sight deflection angle does not exceed the maximum allowable value, avoiding image blurring, and providing accurate quantitative data for subsequent waypoint generation. This is a fundamental prerequisite for route planning in the second region.

[0100] Step B32: Extract the vertices associated with the maximum curvature in the second partition and connect them to form feature ridges; The characteristic ridge line refers to the line formed by sequentially connecting the vertices associated with the maximum curvature within the second zone. It accurately reflects the distribution and curvature trend of the high-curvature area in the second zone and serves as the baseline for route planning in the second zone. As the benchmark for waypoint planning in the second zone, orthogonal slicing along the characteristic ridge line can obtain uniformly distributed surface anchor points, ensuring that waypoints conform to the curvature characteristics of the second zone and avoiding waypoint deviations or blind spots caused by severe surface curvature.

[0101] In one optional approach, the maximum principal curvature data of all vertices in the second partition is first obtained. The maximum principal curvature of all vertices is sorted, and the vertices with the highest maximum principal curvature values ​​(e.g., 30%) are selected. These vertices are the vertices associated with the maximum curvature. Then, for the selected vertices associated with the maximum curvature, vertices with abnormal coordinates are removed to ensure that all vertices are distributed in the area of ​​the second partition. These vertices associated with the maximum curvature are connected in spatial order to fit a smooth curve, which is the feature ridge. After the connection is completed, the feature ridge is smoothed to eliminate the angle jumps during the vertex connection process, ensuring that the feature ridge can accurately fit the bending trend of the high curvature area of ​​the second partition. At the same time, it is checked whether the feature ridge covers the area of ​​the second partition. If there are omissions, the corresponding maximum curvature associated vertices are added, and the ridge is reconnected to form the feature ridge. By extracting characteristic ridge lines that can characterize the distribution and bending trend of high curvature areas in the second partition, a baseline is provided for subsequent waypoint planning. This ensures that the generated waypoints can fit the curved surface of the second partition, avoiding waypoint deviations or blind spots caused by severe curvature of the surface. At the same time, it simplifies the complexity of waypoint planning and improves the accuracy of waypoint planning.

[0102] Step B33: Perform orthogonal slicing along the feature ridge line at the second waypoint step size to obtain surface anchor points. Offset the surface anchor points outward by a preset safe distance along the normal vector of the corresponding vertex to generate waypoints. Orthogonal slicing refers to the process of generating a slicing plane perpendicular to the feature ridge line along its extension direction, and then cutting the 3D surface of the second partition. The slicing plane's perpendicularity to the feature ridge line ensures that the slices uniformly cover the areas on both sides of the ridge line. By obtaining uniformly distributed surface anchor points within the second partition, it is ensured that the surface anchor points conform to the curvature trend of the feature ridge line, covering the high-curvature area of ​​the second partition. This provides accurate reference points for waypoint generation, ensuring that the waypoint distribution adapts to the curvature characteristics of the surface.

[0103] In one optional approach, firstly, the extension direction of the feature ridge is determined. Along the extension direction of the feature ridge, a series of slice planes perpendicular to the feature ridge are generated according to a determined second waypoint step size. The distance between each slice plane is equal to the second waypoint step size, ensuring that the slice planes uniformly cover the entire feature ridge. Then, each slice plane is intersected with the 3D surface of the second partition to obtain the intersection line of each slice plane and the 3D surface. A preset number of key points are extracted from each intersection line; these key points are the surface anchor points, ensuring that each surface anchor point is located on the 3D surface of the second partition and is uniformly distributed. Next, each... The normal vector of the blade vertex corresponding to the location of each surface anchor point is determined, ensuring that the normal vector is a unit vector perpendicular to the blade surface and pointing outwards. The 3D coordinates of each surface anchor point are multiplied by the normal vector to obtain an offset vector, the length of which is equal to the preset safe flight distance. The 3D coordinates of each surface anchor point are added to the corresponding offset vector to obtain new 3D coordinates, which are the waypoints. After generating all waypoints, each waypoint is verified to determine whether the distance between the waypoint and the blade surface is equal to the preset safe flight distance. If not, the direction of the normal vector or the offset distance is adjusted to ensure that all waypoints meet the safe flight requirements.

[0104] Step B34: Generate the route corresponding to the second partition based on the waypoints corresponding to each surface anchor point.

[0105] In one optional approach, all generated waypoints are first sorted by spatial coordinates. Following the extension direction of the feature ridge, the waypoints are arranged sequentially from one end of the feature ridge to the other, ensuring that waypoints on the same slice plane are arranged in chordal order to avoid flight instability caused by waypoint misalignment. Then, curve fitting is performed on the sorted waypoints to eliminate angle jumps and distance abrupt changes between adjacent waypoints, making the transition between waypoints smoother and adapting to the flight characteristics of the UAV. Finally, the smoothed waypoints are connected sequentially to form a complete second-zone route. The route is then checked for integrity to ensure that it covers the entire surface of the second zone without blind spots. The curvature of the route is also checked to ensure it is suitable for the flight capabilities of the UAV. If there are sections with excessive flight difficulty, the coordinates of the corresponding waypoints are adjusted and reconnected to form a new route.

[0106] In this embodiment, the second waypoint step size is determined by the maximum permissible line-of-sight deflection angle and radius of curvature, and the positive correlation between the two is clearly established. This ensures that the waypoint step size can accurately adapt to the curvature of different parts of the second partition. The more severe the surface curvature (the smaller the radius of curvature), the smaller the waypoint step size and the greater the waypoint density. This effectively avoids the problems of insufficient inspection accuracy in high curvature areas or redundancy in smooth transition areas caused by the traditional uniform waypoint step size, thus balancing inspection accuracy and efficiency. Secondly, by extracting the vertices associated with the maximum curvature and connecting them to form feature ridges, the distribution and curvature trend of the high curvature area in the second partition are accurately captured, providing a reliable benchmark for waypoint planning. This ensures that the subsequently generated waypoints can fit the curved surface of the blade, avoiding inspection blind spots caused by severe surface curvature and improving the accuracy of route planning. Secondly, surface anchor points are obtained by orthogonally slicing along the feature ridge and then offset to generate waypoints. This ensures a safe distance between the waypoints and the blade surface, avoiding the risk of UAV collisions, while also ensuring that the waypoints uniformly cover the high curvature area of ​​the second partition. This ensures that the camera shooting angle always meets the requirements and can clearly capture the details of defects on the blade surface, meeting the needs of high-precision inspection. Finally, a complete flight path is generated, ensuring that the path is stable and executable, adapted to the flight characteristics of the UAV, and solving the problems of instability or insufficient inspection accuracy of traditional flight paths in high curvature areas. This achieves refined and high-precision inspection of the second partition.

[0107] When the partition is the third partition, the above step S30 includes steps C31 to C34: Step C31: Extract the extreme points of the spanwise coordinates of the wind turbine blades as the tip sphere center and establish a local spherical coordinate system; The extreme point of the spanwise coordinates refers to the point where the spatial coordinate value of the wind turbine blade is the largest along the spanwise main axis, that is, the very tip of the blade. This point is the reference point of the blade tip and can accurately locate the spatial position of the blade tip.

[0108] The blade tip center of gravity (SOC) refers to the point centered on the extreme point of the spanwise coordinates of the wind turbine blade. It is the origin of the local spherical coordinate system, and its spatial coordinates are completely consistent with those of the extreme point of the spanwise coordinates. It is used to establish a local spherical coordinate system around the blade tip. As the origin of the local spherical coordinate system, it provides a spatial reference for the subsequent establishment of the hemispherical virtual envelope and the generation of waypoints, ensuring that all waypoints are distributed around the blade tip, fully covering the third zone, and avoiding blind spots in inspection.

[0109] A local spherical coordinate system refers to a spherical coordinate system established with the blade tip center as the origin to describe the spatial position of the third zone. This coordinate system uses three parameters—azimuth, polar angle, and radius—to characterize the position of spatial points. It can accurately describe the spatial morphology of the endpoint miniature area and adapt to the contraction characteristics of the third zone. For example, with the blade tip center as the origin and the shortest distance from the center to the blade tip surface set as the reference radius, the azimuth is distributed along the circumference of the blade tip, and the polar angle is distributed along the contraction direction of the blade tip. By combining the azimuth and polar angles, every position on the blade tip surface can be accurately located, providing coordinate support for waypoint generation.

[0110] In one alternative approach, the coordinates of all vertices of the three-dimensional model of the wind turbine blade are first obtained. The vertex with the largest coordinate value along the spanwise principal axis is selected. This vertex is the extreme point of the spanwise coordinates, which is the very tip of the blade. The three-dimensional spatial coordinates (X0, Y0, Z0) of this vertex are recorded, and these coordinates are determined as the coordinates of the blade tip center. Then, a local spherical coordinate system is established with the blade tip center as the origin. The three parameters of the coordinate system are set: azimuth, polar angle, and radius. The azimuth is the rotation angle around the vertical axis of the blade tip center, ranging from 0 degrees to 360 degrees. The polar angle is the angle between the vector from the blade tip center to the waypoint and the blade tip axis, ranging from 0 degrees to 90 degrees. The radius is the distance from the blade tip center to the waypoint, which is the close-up radius derived later.

[0111] Step C32: Based on the target ground sampling distance, camera pixel size and focal length, the close-up radius is derived in reverse, and a hemispherical virtual envelope with a radius equal to the close-up radius is constructed; The target ground sampling distance refers to the pre-set actual ground distance corresponding to each pixel in the image captured by the UAV camera on the blade surface. Its value is determined based on the inspection accuracy requirements of the third zone and is a parameter for deriving the close-up radius. As a constraint condition for deriving the close-up radius, it ensures that the derived close-up radius can make the image captured by the camera meet the preset inspection accuracy requirements, and that the actual distance corresponding to the image pixel does not exceed the target ground sampling distance, ensuring that the defect details of the blade tip can be clearly captured.

[0112] Camera pixel size refers to the actual physical size of each pixel in the camera sensor on a UAV, i.e., the length and width of the pixel. It is a parameter characterizing the camera's imaging accuracy; the smaller the value, the higher the imaging accuracy and the richer the details captured. Combined with the target ground sampling distance and the camera's focal length, the close-up radius is derived in reverse. It is one of the parameters for calculating the close-up radius, ensuring that the derived close-up radius is compatible with the camera's imaging characteristics, so that the captured images meet the inspection accuracy requirements.

[0113] The close-up radius is a radius derived by inversely from the target ground sampling distance, camera pixel size, and focal length. It is a parameter used to construct a hemispherical virtual envelope, and its value determines the size of the hemispherical virtual envelope and the distance between waypoints and the center of the leaf tip. For example, with a target ground sampling distance of 0.1 mm, a camera pixel size of 2.4 μm, and a focal length of 10 mm, the close-up radius is derived to be 0.2 m. A hemispherical virtual envelope with a radius of 0.2 m and centered on the center of the leaf tip is constructed, and all waypoints are distributed on this envelope.

[0114] The hemispherical virtual envelope surface refers to a hemispherical virtual surface built around the blade tip with the tip center as the center and the radius as the close-up radius. This surface can completely enclose the entire surface of the third zone and serves as the spatial carrier for waypoint generation, adapting to the miniaturized and contracted geometry of the blade tip. It provides a uniform spatial carrier for waypoint generation, ensuring that waypoints are evenly distributed around the blade tip, fully covering the third zone, avoiding blind spots in inspection, and providing a basis for the application of spherical azimuth and polar angle steps, simplifying the waypoint generation process.

[0115] In one optional approach, the preset target ground sampling distance, UAV camera pixel size, and focal length parameters are first retrieved. A reverse-engineered formula is used to calculate the close-up radius, which is calculated as (focal length * pixel size) / target ground sampling distance. This formula yields a specific value for the close-up radius, ensuring that the image captured by the camera meets the preset inspection accuracy requirements, i.e., the actual ground distance corresponding to each pixel in the image equals the target ground sampling distance. After calculation, the close-up radius is verified to ensure its value is greater than the minimum safe distance between the UAV and the blade surface, avoiding the risk of collision due to waypoints being too close to the blade tip. Then, a hemispherical virtual envelope is constructed with the blade tip center as the center and the derived close-up radius as the radius. The hemisphere of this envelope faces one side of the blade body, ensuring that the envelope completely covers the entire surface of the third section without any omissions. After construction, the fit between the envelope and the third section is verified. If the envelope is too large or too small, the target ground sampling distance is adjusted, the close-up radius is re-derived, and the envelope is reconstructed.

[0116] Step C33: Calculate the spherical azimuth and polar angle step size based on the preset image overlap rate, and generate waypoints on the hemispherical envelope surface; The preset image overlap rate refers to the minimum overlap ratio between two adjacent inspection images to ensure that the inspection images of the third zone can be completely stitched together without inspection blind spots. Its value is determined according to the inspection accuracy requirements of the third zone and is an important constraint condition for calculating the spherical azimuth angle and polar angle step size.

[0117] Spherical azimuth refers to the angular parameter used in a local spherical coordinate system to describe the horizontal distribution of waypoints (around the leaf tip center). It is measured along the horizontal direction of the hemispherical virtual envelope with the leaf tip center as the origin, and ranges from 0 degrees to 360 degrees. It is one of the parameters for waypoint positioning.

[0118] Polar step size refers to the angular parameter used in a local spherical coordinate system to describe the vertical distribution of waypoints (along the blade tip contraction direction). It is measured with the blade tip center as the origin and along the vertical direction of the hemispherical virtual envelope surface, ranging from 0 degrees to 90 degrees. It is one of the parameters for waypoint positioning.

[0119] In one optional approach, a preset image overlap rate is first retrieved. Then, combining the close-up radius of the hemispherical virtual envelope and the camera's field of view, the spherical azimuth step and polar angle step are calculated. When calculating the spherical azimuth step, the camera's horizontal field of view is used as a basis, combined with the preset image overlap rate, and the formula is used: azimuth step equals camera's horizontal field of view multiplied by (1 minus the preset image overlap rate), ensuring that images captured by adjacent waypoints in the horizontal direction meet the overlap requirements. When calculating the polar angle step, the camera's vertical field of view is used as a basis, combined with the preset image overlap rate, and the formula is used: polar angle step equals camera's vertical field of view multiplied by (1 minus the preset image overlap rate), ensuring that images captured by adjacent waypoints in the vertical direction meet the overlap requirements. The overlap requirement is met. After the calculation is completed, the two step sizes are checked to ensure that the step size meets the flight control accuracy of the UAV. If it does not meet the requirement, the preset image overlap rate is adjusted and the calculation is repeated until the requirement is met. Then, on the hemispherical virtual envelope, waypoints are generated according to the calculated spherical azimuth step size and polar angle step size. Starting from azimuth 0 degrees and polar angle 0 degrees, the azimuth step size is increased sequentially to generate waypoints at the same polar angle level. After completing one polar angle level, the polar angle step size is increased to enter the next polar angle level until the polar angle reaches 90 degrees, generating all waypoints. After generating waypoints, the coordinates of each waypoint are checked to ensure that the waypoints are all located on the hemispherical virtual envelope and are evenly distributed without overlap or omission.

[0120] Step C34: Generate the routes corresponding to the third zone based on each waypoint.

[0121] In one optional approach, all generated waypoints are sorted according to polar angles from 0 to 90 degrees, and within each polar angle level, they are arranged in order of azimuth from 0 to 360 degrees. This ensures that the waypoint order conforms to the UAV's flight logic and avoids flight instability caused by waypoint misalignment. Then, a smoothing algorithm is used to perform curve fitting on the sorted waypoints, eliminating angle jumps and distance abrupt changes between adjacent waypoints, making the transitions between waypoints smoother and adapting to the UAV's flight characteristics. In particular, the smoothing intensity is appropriately increased at sharp blade tips to ensure stable UAV flight. Finally, the smoothed waypoints are connected in sequence to form a complete third-zone flight path. The flight path is then checked for integrity to ensure that it covers the entire surface of the third zone without any blind spots. At the same time, the flight path is checked to ensure that it conforms to the UAV's flight capabilities. If there are sections with excessive flight difficulty, the azimuth and polar angles of the corresponding waypoints are adjusted, and the paths are reconnected to form the third-zone flight path.

[0122] In this embodiment, by extracting the extreme point of the spanwise coordinates as the center of the blade tip, a local spherical coordinate system is established, which accurately locates the spatial position of the blade tip. This adapts to the miniaturized geometric characteristics of the blade tip and solves the problem that traditional coordinate systems cannot accurately describe the spatial morphology of the blade tip. This provides a reliable spatial reference for subsequent waypoint planning, ensuring that waypoints can be distributed around the blade tip. Secondly, based on the target ground sampling distance, camera pixel size, and focal length, the close-up radius is derived inversely, and a hemispherical virtual envelope is constructed. This ensures that the images captured by the camera can meet the preset inspection accuracy requirements and clearly capture the minute defects at the blade tip. It also ensures that the envelope can completely cover the third zone, avoiding blind spots in the inspection. At the same time, it takes into account the flight safety of the UAV and avoids collisions caused by waypoints being too close to the blade tip. Secondly, based on the preset image overlap rate, the spherical azimuth and polar angle step size are calculated to generate uniformly distributed waypoints on the envelope surface. This ensures that images captured by adjacent waypoints meet the overlap requirements, avoiding image stitching failures or blind spots due to insufficient image overlap. Simultaneously, the uniform distribution of waypoints ensures precise coverage of every area at the blade tip, enabling multi-view inspection. Finally, by generating complete flight paths, the system ensures smooth and executable routes, adapting to the flight characteristics of UAVs. This solves the problems of traditional flight paths' difficulty in covering the blade tip and unstable flight, achieving full coverage and high-precision inspection of the miniaturized endpoint area of ​​the third zone. This fills the gap in blade tip inspection and, in conjunction with the flight path planning strategies of the second and first zones, forms a complete wind turbine blade inspection flight path system. This effectively improves the quality and efficiency of the entire wind turbine blade inspection, ensuring precise inspection of every part of the blade, timely detection of defects, and guaranteeing the safe and stable operation of the wind turbine.

[0123] Furthermore, the above methods also include: Step S110: Control the camera principal optical axis vector at each waypoint in the first partition to keep it parallel to the normal vector of the corresponding vertex, and inversely calculate the camera principal optical axis vector into the pitch angle and yaw angle of the camera gimbal. The camera's principal optical axis vector is the vector pointing from the center of the camera lens to the target (the vertex corresponding to the blade surface). It represents the camera's shooting direction and is a parameter that controls the camera's shooting angle. Its direction directly determines the camera's angle of view when shooting the blade. For example, if the camera's principal optical axis vector is a straight line pointing from the camera lens to a vertex of the blade, and this vector is parallel to the vertex's normal vector, then the camera lens is directly facing the blade surface where that vertex is located, and a clear frontal image of that area can be captured.

[0124] Antiparallelism means that the directions of the two vectors are completely opposite and they are always on the same straight line. That is, the angle between the camera's principal optical axis vector and the normal vector of the corresponding vertex is 180 degrees and they are in opposite directions. This is the control requirement for the principal optical axis vector of the camera in the first partition.

[0125] The pitch angle refers to the angle of rotation of the camera gimbal in the vertical direction, that is, the angle between the camera's main optical axis and the horizontal plane. It is used to adjust the shooting angle of the camera in the vertical direction and is one of the parameters for calculating the camera's attitude.

[0126] Yaw angle refers to the angle of rotation of the camera gimbal in the horizontal direction, that is, the angle between the camera's main optical axis and the drone's flight direction. It is used to adjust the camera's shooting angle in the horizontal direction and is one of the parameters for calculating the camera's attitude.

[0127] In one optional approach, the three-dimensional spatial coordinates of each waypoint in the first partition and the normal vector of the corresponding blade vertex are first obtained, ensuring that the normal vector is a unit vector perpendicular to the blade surface and pointing outwards. Then, by calling the camera attitude control module, the direction of the camera's principal optical axis vector is adjusted in real time and compared with the normal vector of the corresponding vertex. If the angle between the two is not 180 degrees, the camera gimbal is driven to rotate, gradually adjusting the direction of the principal optical axis vector until the principal optical axis vector and the normal vector are completely opposite in direction and on the same straight line, achieving reverse parallel control. After the principal optical axis vector adjustment is completed, the three-dimensional coordinate parameters of the current principal optical axis vector are obtained, combined with the UAV's flight attitude. For parameters such as heading angle, pitch angle, roll angle, and waypoint coordinates, the principal optical axis vector is inversely calculated into the camera gimbal's pitch and yaw angles. The specific calculation process is as follows: using the UAV's body coordinate system as a reference, the principal optical axis vector is decomposed into vertical and horizontal components. The vertical component corresponds to the gimbal's pitch angle, and the horizontal component corresponds to the gimbal's yaw angle. The specific values ​​of the two angles are calculated using trigonometric functions. After the calculation is completed, the pitch and yaw angle commands are sent to the camera gimbal to drive the gimbal to adjust to the corresponding angles, ensuring that the camera's principal optical axis vector always remains parallel to the corresponding vertex normal vector. At the same time, the adjustment process for each waypoint is checked in real time, and any deviation is corrected promptly.

[0128] Understandably, the above method ensures that the camera at each waypoint in the first zone can capture a frontal view of the blade surface, minimizing image distortion and obtaining clear and complete frontal inspection images. At the same time, by reverse-engineering the gimbal angle, precise execution commands are provided for camera attitude control, ensuring the accuracy and stability of the camera's shooting angle, adapting to the smooth geometric characteristics of the first zone, improving the inspection image quality of the first zone, and providing reliable image support for subsequent defect identification.

[0129] Step S120: Control the principal optical axis vector of the camera at each waypoint in the second partition to be orthogonal to the normal vector of the corresponding vertex, and inversely calculate the camera principal optical axis vector into the pitch angle and yaw angle of the camera gimbal. Orthogonality refers to two vectors being perpendicular to each other, meaning the angle between the camera's principal optical axis vector and the normal vector of the corresponding vertex is 90 degrees, and they are not on the same straight line. This is a control requirement for the principal optical axis vector of the camera in the second section. For example, if the normal vector of the corresponding vertex is perpendicular to the blade surface outwards, then the camera's principal optical axis vector must be perpendicular to this normal vector direction, that is, parallel to the tangent direction of the blade surface, so that the curved part can be photographed from the side, clearly showing the curvature and surface details of the surface.

[0130] In one optional approach, the three-dimensional spatial coordinates of each waypoint in the second partition and the normal vector of the corresponding blade vertex are first obtained. Due to the severe curvature of the surface in the second partition, the normal vector of the corresponding vertex needs to be updated in real time to ensure that the direction of the normal vector is consistent with the actual orientation of the blade surface. Then, the camera attitude control module is activated to adjust the direction of the camera's principal optical axis vector and calculate the angle between it and the normal vector of the corresponding vertex. If the angle is not 90 degrees, the camera gimbal is driven to perform pitch and yaw movements, gradually adjusting the principal optical axis vector until it is perpendicular to the normal vector, achieving orthogonal control. At the same time, it is ensured that the principal optical axis vector points towards the blade surface and avoids pointing outwards from the blade. The principal optical axis vector is then completed. After adjustment, the three-dimensional coordinates of the current principal optical axis vector are obtained. Combined with the UAV's flight attitude and waypoint coordinates, the same coordinate transformation algorithm as the first partition is used to inversely calculate the principal optical axis vector into the camera gimbal's pitch and yaw angles. During the calculation process, the dynamic changes in the normal vector direction of the second partition need to be considered. The calculation results for each waypoint are checked separately to avoid angle calculation errors caused by changes in the normal vector. The calculated pitch and yaw angle commands are sent to the gimbal to drive the gimbal to adjust in real time, ensuring that the camera's principal optical axis vector at each waypoint is orthogonal to the corresponding vertex normal vector. At the same time, the principal optical axis direction is monitored in real time, and if any deviation occurs, it is corrected immediately to ensure that the shooting angle always meets the requirements.

[0131] Understandably, by adapting to the severe curvature of the second zone's surface using the above method, the camera can capture images of the curved parts of the blades from the side, avoiding image distortion and loss of detail when shooting from the front. This allows for the clear capture of defects such as cracks and wear at the curved parts of the blade's leading and trailing edges. At the same time, by accurately calculating the gimbal angle, the camera's posture can be dynamically adjusted, ensuring the clarity and integrity of the second zone's inspection images. This provides high-quality image data for defect identification and improves the inspection accuracy of the second zone.

[0132] Step S130: Control the main optical axis of the camera at each waypoint in the third partition to point towards the center of the leaf tip.

[0133] In one alternative approach, the three-dimensional spatial coordinates of the leaf tip center of the third partition and the three-dimensional spatial coordinates of each waypoint in the third partition are first obtained, ensuring that both coordinates are accurate. Then, the camera attitude control module is activated to calculate in real time the vector pointing from the camera lens center to the leaf tip center at each waypoint, and this vector is determined as the target direction of the camera's principal optical axis. Through the UAV flight control system, the camera gimbal is driven to perform pitch and yaw movements to adjust the direction of the camera's principal optical axis vector, which is then compared with the target vector. If the two directions are inconsistent, the gimbal angle is gradually adjusted until the camera's principal optical axis is aligned correctly. The optical axis vector and the target vector are completely coincident, meaning the principal optical axis points precisely to the center of the leaf tip. Since the waypoints in the third section are distributed around the leaf tip, the principal optical axis direction corresponding to each waypoint is different. Each waypoint needs to be adjusted and calculated individually, and the target direction of the principal optical axis is updated in real time to ensure that the camera's principal optical axis always points to the center of the leaf tip, no matter which waypoint the UAV flies to. At the same time, the pointing accuracy of the principal optical axis is checked in real time. If a pointing deviation occurs, that is, if the deviation exceeds the preset threshold, the gimbal angle is adjusted immediately to ensure that the pointing accuracy meets the inspection requirements and avoids the leaf tip inspection blind spot caused by pointing deviation.

[0134] Understandably, by adapting to the miniaturized and contracted geometric characteristics of the leaf tip in the third section, the camera can fully cover the leaf tip from multiple perspectives. No matter where the drone flies around the leaf tip, it can accurately aim at the center of the leaf tip, avoiding blind spots caused by the small size and special shape of the leaf tip. At the same time, it ensures that all areas such as the end face and side of the leaf tip can be clearly photographed, capturing tiny defects in the leaf tip and improving the comprehensiveness and accuracy of the inspection in the third section.

[0135] In this embodiment, considering the gentle geometry of the first zone, the camera's principal optical axis is controlled to be parallel to the corresponding vertex normal vector, enabling frontal shooting and minimizing image distortion. This ensures that details on the blade surface of the first zone are captured clearly and completely. Simultaneously, by inversely calculating the gimbal angle, precise control of the camera's attitude is achieved, ensuring the stability and accuracy of the shooting angle and improving the quality and consistency of the inspection images for the first zone. Secondly, considering the high curvature and severe surface bending of the second zone, the camera's principal optical axis is controlled to be orthogonal to the corresponding vertex normal vector, enabling side shooting. This avoids image distortion and detail loss caused by surface convexities or concaveities during frontal shooting, accurately capturing defects such as cracks and wear at the leading and trailing edges of the blades. Furthermore, by dynamically calculating the gimbal angle to adapt to dynamic changes in the normal vector direction, the shooting angle at each waypoint meets the requirements, significantly improving the inspection accuracy of the second zone. Secondly, considering the miniaturized and contracted nature of the blade tip in the third section, the camera's main optical axis is always pointed towards the center of the blade tip, achieving multi-angle, full-coverage imaging of the blade tip area. This solves the inspection blind spot problem caused by the small size and unique shape of the blade tip, ensuring that all areas, including the end face and sides of the blade tip, can be clearly captured, capturing minute defects and filling the gap in blade tip inspection. Finally, by employing differentiated camera control strategies for the three sections, refined and differentiated control of the camera attitude is achieved. Combined with the section flight path planning strategy, this forms a complete inspection solution with section flight paths and section camera control, ensuring that every area of ​​the entire wind turbine blade can be captured at the optimal angle, obtaining high-quality inspection images.

[0136] In one feasible implementation, please refer to Figure 3 Step S40 also includes: Step S41: Concatenate the routes corresponding to each partition to obtain the initial route; The initial flight path refers to the UAV flight path obtained by directly splicing the flight paths corresponding to each zone. It has not undergone smoothing, resampling, and filtering processing, and only achieves a simple connection between the flight paths of each zone. It may have problems such as abrupt trajectory changes at the boundary and uneven distribution of waypoints. For example, directly splicing the flight paths of the first, second, and third zones in the order of blade spanwise results in an initial flight path that includes waypoints of all zones. This flight path can cover the entire wind turbine blade, but there may be problems such as abrupt waypoint connections and discontinuous trajectories at the boundary of zones.

[0137] In one optional approach, the flight path data corresponding to each zone is first acquired, including the three-dimensional coordinates of all waypoints in each zone's flight path, the camera gimbal pitch and yaw angles associated with the waypoints, and the flight sequence of each zone's flight path. Following the spanwise sequence of the wind turbine blades, the flight paths of the first, second, and third zones are sequentially stitched together, ensuring that the end waypoint of the previous zone's flight path is spatially adjacent to the starting waypoint of the next zone's flight path. During the stitching process, the original waypoint sequence and camera gimbal parameters of each zone's flight path are preserved, without altering the waypoint distribution and attitude within each zone's flight path. Only the beginning and end of each zone's flight path are connected to form an initial flight path containing all zone waypoints. By integrating the independent flight paths of each zone into a complete initial flight path, full coverage of all zone inspection areas is achieved, ensuring that the UAV can complete the entire wind turbine blade inspection task through a single flight path. This also provides a foundation for subsequent smoothing, resampling, and filtering processes, avoiding inspection omissions caused by the independence of zone flight paths.

[0138] Step S42: Smooth the intersection of the routes in each partition to obtain the transition trajectory corresponding to the intersection. The boundary between two adjacent flight paths refers to the connecting area between flight paths corresponding to two adjacent flight paths. It is a transitional area for waypoints between the two flight paths. The waypoints in this area belong to the two adjacent flight paths and have differences in trajectory direction, waypoint density, etc., which can easily lead to abrupt trajectory changes. For example, the area between the terminal waypoint of the flight path in the first flight path and the starting waypoint of the flight path in the second flight path is the boundary between flight paths. Waypoints in this area may have abrupt coordinate changes or excessive changes in camera gimbal angle, which require smoothing.

[0139] A transition trajectory refers to a smooth trajectory obtained at the boundary of two adjacent zone routes after smoothing processing. Its two ends connect to the routes of the two zones respectively, achieving a smooth transition between the two zones without significant abrupt changes. For example, at the boundary between the first and second zones, an arc-shaped transition trajectory is obtained through smoothing. The starting end of this transition trajectory connects to the end of the first zone route, and the ending end connects to the starting end of the second zone route, achieving a smooth transition between the two zones.

[0140] In one alternative approach, the boundary of each zone's flight path is first located, and the end and start waypoints of adjacent two zone flight paths are determined. Based on these two waypoints, a transition zone is defined. The last three waypoints of the preceding zone and the first three waypoints of the following zone within the transition zone are extracted, totaling six waypoints, as the base waypoints for smoothing. A linear interpolation algorithm is used to insert several transition points between these six waypoints. The coordinates of the transition points and the camera gimbal angle are adjusted so that the coordinates and attitude of the transition points can smoothly connect the waypoints of the preceding and following zones. The base waypoints and transition points are fitted together to form a smooth curve, which is the transition trajectory at the boundary. After fitting, the smoothness of the transition trajectory is checked. If there are abrupt changes, the interpolation parameters are adjusted and the curve is refitted until there are no obvious abrupt changes in the transition trajectory.

[0141] Step S43: Based on the control frequency of the UAV, resample the transition trajectory to obtain transition waypoints, and use the transition waypoints to update the initial route. The UAV control frequency refers to the response frequency of the UAV flight control system to waypoint commands, that is, the number of times the UAV can receive and execute waypoint commands per unit time. It determines the density of waypoints that the UAV can stably track and serves as a reference for resampling. For example, if a UAV's control frequency is 10Hz, it can receive and execute 10 waypoint commands per second. During resampling, it is necessary to ensure that the interval between transition waypoints allows the UAV to complete 10 waypoint switching per second, ensuring stable flight.

[0142] Transition waypoints are waypoints obtained by resampling the transition trajectory. They are located on the transition trajectory and their coordinates and associated camera gimbal angles are adapted to the smooth characteristics of the transition trajectory. They are used to replace the harsh waypoints at the boundaries of the partitions in the initial route and realize route updates.

[0143] In one alternative approach, the control frequency parameters of the UAV are first obtained to determine the number of times the UAV can receive and execute waypoint commands per unit time. The resampling time interval is calculated based on the control frequency, i.e., the resampling interval equals 1 divided by the control frequency. Based on the length of the transition trajectory and the resampling interval, the number of transition waypoints to be generated is calculated to ensure that the transition waypoints are evenly distributed on the transition trajectory. Starting from the beginning of the transition trajectory, transition waypoints are selected sequentially according to the resampling interval, and the three-dimensional coordinates and associated camera gimbal angle of each transition waypoint are recorded. The selected transition waypoints are arranged in order to replace the original waypoints at the boundaries of the partitions in the initial route, thus completing the update of the initial route.

[0144] Step S44: Input the updated flight path into the dynamics constraint module, and filter the updated flight path through the dynamics constraint module to obtain the target flight path corresponding to the UAV.

[0145] The dynamics constraint module is a functional module used to filter the UAV flight path and verify the compliance of waypoints. It has built-in dynamic parameter limits for UAV flight, such as maximum angular velocity and maximum acceleration and deceleration, and can perform threshold verification and filtering optimization of waypoints in the flight path.

[0146] In one optional approach, the updated flight path is first fully input into the dynamics constraint module. This module pre-constructs dynamics constraint parameters for UAV flight, including the maximum yaw rate and maximum pitch rate of the camera gimbal, as well as the maximum acceleration and maximum deceleration of the UAV. The dynamics constraint module verifies each waypoint in the updated flight path, calculating the yaw rate and pitch rate of the camera gimbal corresponding to each waypoint, and the acceleration and deceleration of the UAV when it reaches that waypoint. The calculated parameters are compared with preset maximum allowable thresholds. If the parameters of a waypoint exceed the threshold, the waypoint is marked as a non-compliant waypoint. After removing all non-compliant waypoints, the remaining compliant waypoints are obtained. The compliant waypoints are reconnected according to the original flight path order to form a preliminary target flight path. The preliminary target flight path is smoothed to eliminate minor abrupt changes at the connection points of compliant waypoints, resulting in the final UAV target flight path. Finally, the completeness and feasibility of the target flight path are verified to ensure that no inspection area is missed and that it meets the UAV flight requirements. By filtering through the dynamics constraint module, non-compliant waypoints in the updated route are eliminated, optimizing the smoothness and feasibility of the route. This ensures that the target route is compatible with the dynamic characteristics of the UAV, avoiding flight accidents caused by unreasonable waypoint parameters. Ultimately, a safe, stable, and executable UAV target route is obtained, providing reliable path guidance for UAVs to perform inspection tasks.

[0147] In this embodiment, an initial route is obtained by splicing the routes corresponding to each partition, achieving full coverage of all partition inspection areas. Independent partition routes are integrated into a complete inspection path, avoiding fragmentation of partition inspections and ensuring that the UAV can complete the inspection of the entire wind turbine blade via a single route, thus improving inspection efficiency. Secondly, addressing the issue of abrupt trajectory changes at the boundaries of partition routes, a smoothing process is used to generate transition trajectories, and then transition waypoints are generated through resampling. This effectively eliminates abrupt transitions at the boundaries, enabling the UAV to fly smoothly during partition switching and avoiding instability such as sudden acceleration / deceleration and sharp turns, ensuring UAV flight safety and improving the continuity and executability of the route. Thirdly, resampling is performed based on the UAV control frequency to ensure that the distribution density of transition waypoints matches the UAV's response capability. This avoids untimely UAV response due to excessively high waypoint density or insufficient trajectory tracking accuracy due to excessively low waypoint density, further optimizing the practicality of the route. Finally, through the filtering process of the dynamic constraint module, non-compliant waypoints are eliminated to ensure that the target route conforms to the dynamic characteristics of the UAV, thus avoiding flight accidents caused by waypoint parameters exceeding the limits of the UAV, and ultimately obtaining a safe, stable, and fully covered UAV target route.

[0148] Furthermore, step S42 above includes: Step S421: At the intersection of routes in each partition, extract a preset number of waypoints to form a transition neighborhood; The transition neighborhood refers to the area defined around the connecting region at the boundary of the two partition routes, containing a number of waypoints. This area is centered on the first and last waypoints of the two adjacent partition routes and includes a predetermined number of adjacent waypoints. It is the region from which samples are extracted for fitting the transition trajectory. For example, at the boundary between the first and second partitions, the three waypoints at the end of the first partition and the three waypoints at the beginning of the second partition are extracted. The spatial range formed by these six waypoints is the transition neighborhood, which is used for fitting the subsequent transition trajectory.

[0149] In one optional approach, the boundary of each zone's flight path is first located, and the end and start waypoints of two adjacent zone flight paths are identified. A transition neighborhood is defined centered on these two waypoints. Based on a preset number (e.g., 6), half of the preset number of waypoints (e.g., 3) at the end of the preceding zone's flight path are extracted. These waypoints must be closest to the end waypoint and evenly distributed in the end region. Simultaneously, the other half of the preset number of waypoints (e.g., 3) at the beginning of the following zone's flight path are extracted. These waypoints must be closest to the starting waypoint and evenly distributed in the starting region. All extracted waypoints are then aggregated, and waypoints with abnormal coordinates or unreasonable attitudes are removed, ensuring that all waypoints belong to the corresponding zone's flight path and are spatially adjacent. The remaining waypoints are numbered and sorted according to their distance from the boundary, from farthest to closest, forming a transition neighborhood. This ensures that the transition neighborhood covers the connecting area between the two zone flight paths, providing sufficient sample waypoints for subsequent fitting of the transition trajectory. By constructing the transition neighborhood required for transition trajectory fitting, sufficient sample waypoints are provided to ensure that the subsequently fitted transition trajectory can accurately adapt to the connection characteristics of the two partition routes. At the same time, the range of smoothing is limited, laying the foundation for transition trajectory fitting and avoiding insufficient transition trajectory fitting accuracy due to insufficient sample waypoints.

[0150] Step S422: Extract the association information of each waypoint in the transition neighborhood to form the state vector of each waypoint. The association information includes the three-dimensional coordinates of the waypoint, the pitch angle and yaw angle of the camera gimbal associated with the waypoint. A state vector is a vector composed of associated information about a waypoint. It integrates the waypoint's spatial location and camera attitude parameters, and can comprehensively and accurately represent the state of a waypoint. It is the data unit for transition trajectory fitting. For example, the state vector of a waypoint contains five parameters: the waypoint's x-coordinate, y-coordinate, z-coordinate, camera gimbal pitch angle, and camera gimbal yaw angle. This vector can completely describe the waypoint's spatial location and camera attitude.

[0151] In one optional approach, each waypoint within the transition neighborhood is traversed first, and the associated information of each waypoint is extracted one by one. The 3D coordinates of the waypoints are directly retrieved from the UAV flight path planning system to ensure the accuracy of the coordinate data. The camera gimbal pitch and yaw angles associated with the waypoints are extracted from the camera attitude control records to ensure consistency with the camera attitude corresponding to the waypoint. The extracted 3D coordinates, pitch angles, and yaw angles of each waypoint are organized and integrated into a vector in a unified order, which is the state vector of the waypoint. During the integration process, it is ensured that the parameter order of each state vector is consistent, namely x-coordinate, y-coordinate, z-coordinate, pitch angle, and yaw angle, to avoid parameter disorder that may lead to subsequent fitting errors. Each state vector is verified to check whether the parameters are complete and whether the values ​​are reasonable. If there are missing or abnormal parameters, the associated information of the corresponding waypoints is re-extracted, and the state vectors are corrected to ensure that the state vectors of all waypoints are complete and accurate.

[0152] Step S423: Apply cubic B-spline basis functions to the state vectors of each waypoint to obtain the transition trajectory.

[0153] Cubic B-spline basis functions are mathematical functions used for curve fitting. They possess continuous and smooth characteristics, enabling the fitting of a smooth curve from multiple sample points. Furthermore, they closely match the distribution trends of these sample points, making them suitable for fitting transition trajectories at the boundaries of regional flight paths. For example, by inputting the state vectors of six waypoints within the transition neighborhood into a cubic B-spline basis function, the function calculates and fits a curve that smoothly connects these six waypoints. This curve represents the transition trajectory, conforming to the spatial distribution of waypoints and achieving a smooth transition in camera attitude.

[0154] In one optional approach, the parameters of the cubic B-spline basis function are first determined, including the node vectors and degree. Based on the number and distribution of waypoints within the transition neighborhood, reasonable node vectors are set to ensure the cubic B-spline basis function can adapt to the distribution trend of waypoints. The state vector of each waypoint is used as a sample point and input into the cubic B-spline basis function. Through the calculation of the basis function, the sample points are fitted to generate a continuous curve. During the fitting process, the smoothing parameters of the cubic B-spline basis function are adjusted to ensure that the fitted curve fits the state vector of each waypoint, while also ensuring the continuity and smoothness of the curve, avoiding broken lines or abrupt changes. After fitting, the obtained curve is verified by calculating the deviation between the curve and the state vector of each waypoint. If the deviation exceeds a preset threshold, the parameters of the cubic B-spline basis function are adjusted, and the fitting is repeated until the deviation is within the allowable range. The final smooth curve is the transition trajectory at the boundary, which can simultaneously match the spatial position of the waypoint and the camera attitude, achieving a smooth connection between the two partition routes. By fitting the cubic B-spline basis function, a smooth and continuous transition trajectory is obtained. This trajectory can accurately connect the flight paths of the two partitions, eliminate abrupt changes in trajectory at the boundary, and achieve a smooth transition of camera attitude. It provides a stable flight path for the UAV, avoids flight instability caused by abrupt changes in trajectory, and provides a reliable trajectory basis for the generation of subsequent transition waypoints.

[0155] In this embodiment, a transition neighborhood is constructed by extracting a preset number of waypoints, ensuring that the sample waypoints required for fitting the transition trajectory are sufficient and representative. This avoids the problem of the transition trajectory not matching the actual connection requirements due to insufficient samples, laying the foundation for high-precision fitting of the transition trajectory. Secondly, by extracting the association information of waypoints to form a state vector, the spatial position of the waypoints and the camera attitude are integrated into a unified fitting unit. This allows the fitted transition trajectory to not only achieve a smooth connection in spatial position but also a smooth transition in camera attitude, avoiding problems such as image blurring and decreased acquisition quality caused by sudden changes in camera attitude during UAV flight, thus improving the consistency of inspection images. Thirdly, a cubic B-spline basis function is used for fitting. Utilizing its continuous and smooth characteristics, the fitted transition trajectory can well fit the distribution trend of waypoints, effectively eliminating trajectory abrupt changes at the boundaries of regional flight paths. This ensures that the UAV can smoothly pass through the boundaries, guaranteeing flight safety and improving the coherence and executability of the flight path.

[0156] Furthermore, step S44 above includes: Step S441: Input the updated flight path into the dynamics constraint module. The dynamics constraint module performs threshold verification on the yaw rate, pitch rate and UAV flight acceleration and deceleration of the camera gimbal associated with each waypoint in the updated flight path. Yaw rate refers to the angular velocity of a camera gimbal in the horizontal direction, that is, the change in the yaw angle of the camera gimbal per unit time. It is used to characterize the speed of the camera gimbal's yaw motion and is one of the parameters for measuring the rate of change of camera attitude. For example, if the maximum permissible yaw rate of a certain camera gimbal is 15 degrees per second, and the yaw rate corresponding to a certain waypoint is 20 degrees per second, exceeding the maximum permissible threshold, then that waypoint is a non-compliant waypoint and needs to be eliminated.

[0157] Pitch rate refers to the angular velocity of a camera gimbal in the vertical direction, that is, the change in the pitch angle of the camera gimbal per unit time. It is used to characterize the speed of the camera gimbal's pitch movement and is one of the parameters for measuring the rate of change of camera attitude. For example, if the maximum permissible pitch rate of a certain camera gimbal is 12 degrees per second, and the pitch rate corresponding to a waypoint is 15 degrees per second, exceeding the maximum permissible threshold, then that waypoint is a non-compliant waypoint and needs to be eliminated.

[0158] Drone flight acceleration and deceleration refer to the rate at which a drone's speed increases or decreases during flight; that is, the change in the drone's flight speed per unit time. It is divided into acceleration and deceleration and is one of the parameters used to measure the stability of drone flight. For example, if a drone's maximum permissible acceleration is 2 meters per second squared and its maximum permissible deceleration is 3 meters per second squared, and if a waypoint has a flight acceleration of 2.5 meters per second squared, exceeding the maximum permissible limit, then that waypoint is a non-compliant waypoint and needs to be removed.

[0159] In one optional approach, the updated flight path is first fully input into the dynamics constraint module. This module pre-constructs threshold parameters such as the maximum permissible yaw rate and maximum permissible pitch rate of the camera gimbal, as well as the maximum permissible acceleration and maximum permissible deceleration of the UAV. Then, each waypoint in the updated flight path is traversed, and the associated yaw rate and pitch rate of the camera gimbal, as well as the change in flight speed between that waypoint and adjacent waypoints, are extracted. The acceleration and deceleration of the UAV when it reaches that waypoint are calculated. Finally, the yaw rate of each waypoint is compared with the maximum permissible yaw rate and maximum permissible pitch rate of the camera gimbal. The system allows for comparison of yaw rate; if the yaw rate exceeds the maximum allowable value, the waypoint is marked as non-compliant. It also compares the pitch rate of each waypoint with the maximum allowable pitch rate; if the pitch rate exceeds the maximum allowable value, the waypoint is marked as non-compliant. Furthermore, it compares the flight acceleration and deceleration of each waypoint with the maximum allowable acceleration and deceleration; if the corresponding limits are exceeded, the waypoint is marked as non-compliant. After threshold verification for all waypoints, a list of all marked non-compliant waypoints is compiled.

[0160] Step S442: Waypoints that exceed the maximum permissible angular velocity and acceleration / deceleration limits are removed to obtain compliant waypoints; The maximum permissible angular velocity refers to the pre-set maximum allowable values ​​for the camera gimbal's yaw and pitch angular velocities. Determined based on the hardware performance of the UAV and camera, it serves as the benchmark for threshold verification, used to determine whether the camera attitude changes corresponding to waypoints are compliant. For example, based on the camera gimbal's hardware performance, the maximum permissible yaw angular velocity is set to 15 degrees per second, and the maximum permissible pitch angular velocity to 12 degrees per second. All waypoint angular velocity parameters must not exceed these limits; otherwise, they are considered non-compliant.

[0161] Compliant waypoints refer to waypoints that, after threshold verification, have yaw rate, pitch rate, and UAV acceleration / deceleration values ​​associated with them not exceeding the maximum allowable limits. Their parameters conform to the dynamic characteristics of the UAV and can be used to construct target flight paths.

[0162] In one optional approach, the list of non-compliant waypoints obtained after threshold verification is first retrieved, and the violation type and location of each non-compliant waypoint are identified. The updated flight path is then traversed, and all non-compliant waypoints in the list are removed one by one, retaining only unmarked waypoints. After removal, the remaining waypoints are reviewed to check for any missed non-compliant waypoints; if any are found, they are removed. Simultaneously, the order of the remaining waypoints is checked to ensure that the order of the remaining waypoints still conforms to the flight logic of the flight path after removing non-compliant waypoints, i.e., arranged according to the zonal span order and the direction of the transition trajectory. The remaining waypoints are renumbered to ensure a continuous order and avoid any issues with disorder. Finally, the remaining waypoints are verified again to confirm that the yaw rate, pitch rate, and flight acceleration / deceleration of all waypoints do not exceed the maximum allowable limits, ensuring that all obtained waypoints are compliant.

[0163] Step S443: Based on compliant waypoints, construct the target flight path for the drone.

[0164] In one optional approach, all compliant waypoints are first sorted according to the UAV's flight logic, i.e., the spanwise sequence from blade root to blade tip, and the direction of the transition trajectory. The waypoints are then rearranged to ensure the order conforms to the UAV's flight path, avoiding flight confusion caused by disordered order. Next, trajectory fitting is performed on the sorted compliant waypoints, using a smoothing algorithm to eliminate minor abrupt changes between adjacent waypoints, making the transitions between waypoints smoother and better suited to the UAV's flight characteristics. After fitting, the resulting flight path is validated for completeness, checking for any omissions in the inspection area. If omissions are found, the corresponding compliant waypoints are added, and the sorting and fitting are repeated. Finally, the complete flight path undergoes dynamic verification, again checking whether the parameters of each waypoint meet the dynamic constraints, ensuring that all waypoints in the flight path are compliant, ultimately yielding the target flight path for the UAV.

[0165] In this embodiment, the dynamic constraint module performs threshold verification on the yaw rate, pitch rate, and UAV acceleration / deceleration of waypoints. This comprehensively identifies non-compliant waypoints in the updated flight path, preventing flight accidents such as UAV flight loss of control, camera image blurring, and equipment damage caused by these waypoints, thus ensuring UAV flight safety and image acquisition quality. Secondly, by eliminating non-compliant waypoints and retaining compliant waypoints, it ensures that the subsequently generated target flight path is fully compatible with the UAV's dynamic characteristics, enabling the UAV to fly smoothly and safely along the path. This avoids flight instability caused by waypoint parameters exceeding limits, improving the executability of the flight path. Thirdly, constructing the target flight path based on compliant waypoints ensures the smoothness and continuity of the target flight path, while also ensuring that the target flight path covers all inspection areas without omission, achieving full coverage inspection of wind turbine blades.

[0166] It should be noted that the above examples are only for understanding this application and do not constitute a limitation on the route planning method of this application. Any simple modifications based on this technical concept are within the protection scope of this application.

[0167] Based on the same inventive concept, this application provides a route planning device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the route planning method in the above embodiments.

[0168] like Figure 4As shown, the flight path planning device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the flight path planning device. The processing unit 1001, the ROM 1002, and the RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. Communication device 1009 allows the flight path planning device to communicate wirelessly or wiredly with other devices to exchange data. Although the figure shows flight path planning devices with various systems, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0169] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0170] The flight path planning equipment provided in this application, employing the flight path planning method described in the above embodiments, can divide wind turbine blades into different zones and determine the flight path corresponding to each zone in a targeted manner, thereby improving the accuracy of the UAV's target flight path. Compared with the prior art, the beneficial effects of the flight path planning equipment provided in this application are the same as those of the flight path planning method provided in the above embodiments, and other technical features of this flight path planning equipment are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0171] Based on the same inventive concept, this application provides a computer-readable storage medium having computer-readable program instructions (i.e., computer programs) stored thereon, which are used to execute the route planning method in the above embodiments.

[0172] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory (EPROM), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, radio frequency (RF), etc., or any suitable combination thereof.

[0173] The aforementioned computer-readable storage medium may be included in the route planning device; or it may exist independently and not be installed in the route planning device.

[0174] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the route planning device, cause the route planning device to implement the steps of the aforementioned route planning method.

[0175] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0176] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0177] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0178] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described route planning method. This program can divide the wind turbine blades into different zones and specifically determine the route corresponding to each zone, thereby improving the accuracy of the UAV's target route. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the route planning method provided in the above embodiments, and will not be repeated here.

[0179] The above are only some embodiments of this application and do not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A route planning method characterized by, The method includes: Determine the geometric characteristics of the wind turbine blades; Based on the aforementioned geometric features, the wind turbine blades are divided into different partitions, wherein each partition corresponds to a different part of the wind turbine blades; Determine the appropriate route planning strategy for each partition, and perform route planning for the corresponding partitions based on the route planning strategy to obtain the routes corresponding to each partition; By splicing together the flight paths corresponding to each partition, the target flight path corresponding to the UAV is obtained, wherein the target flight path is the flight path of the UAV when inspecting the wind turbine blades.

2. The route planning method of claim 1, wherein, The determination of the geometric features of the wind turbine blades includes: A three-dimensional model of the wind turbine blade is obtained, and the curvature of the vertices of each triangular facet is determined, wherein the three-dimensional model is composed of multiple triangular facets; Determine the chord length of each section of the wind turbine blades; The curvature of the vertex and the chord length of the cross section are used to determine the geometric features of the wind turbine blade.

3. The route planning method as described in claim 2, characterized in that, The determination of the curvature of the vertices of each triangular facet includes: Traverse each vertex of each of the aforementioned triangular facets, determine the normal vector of each vertex, and establish a local tangent plane coordinate system with the vertex as the origin and the normal vector of the vertex; Search for the neighboring vertices of the vertex centered on the vertex, and project all the neighboring vertices of the vertex onto the local tangent plane coordinate system; In the local tangent plane coordinate system, the quadratic surface equation is fitted by the least squares method based on the projected coordinates of all neighboring vertices of the vertex, and the coefficients of the quadratic term of the quadratic surface equation are extracted. Construct the Hessian matrix based on the coefficients of the quadratic term, and perform eigenvalue decomposition on the Hessian matrix to obtain the curvature of the vertex, wherein the Hessian matrix characterizes the degree of curvature of the surface at the vertex.

4. The route planning method as described in claim 2, characterized in that, Determining the chord length of each section of the wind turbine blade includes: Determine the spanwise principal axis of the wind turbine blade, and generate multiple slice planes orthogonal to the spanwise principal axis along the spanwise principal axis at a preset step size; The intersection of each slice plane with the three-dimensional model of the wind turbine blade is obtained to obtain the cross section at each slice position. The cross section is composed of the set of intersection points between the corresponding slice plane and the three-dimensional model. Each of the cross sections is subjected to convex hull processing to obtain the convex hull contour of each cross section; For each convex hull profile, determine the distance between every two vertices on the convex hull profile, and determine the chord length of the cross section corresponding to the convex hull profile based on the two vertices with the largest distance.

5. The route planning method according to any one of claims 1 to 4, characterized in that, The geometric features include: the curvature of the vertices of each triangular facet of the three-dimensional model of the wind turbine blade, and the chord length of each section of the wind turbine blade; The process of dividing the wind turbine blades into different zones based on the geometric features includes: Based on the conditions satisfied by the aforementioned geometric features, the wind turbine blades are divided into different zones; The partitions include a first partition, a second partition, and a third partition. The geometric features of the first partition satisfy the following conditions: the absolute value of the curvature of the vertex is less than a first preset value, and the chord length of the cross section is greater than a preset chord length. The geometric features of the second partition satisfy the following conditions: the absolute value of the curvature of the vertex is greater than a second preset value. The geometric features of the third partition satisfy the following conditions: the chord length of the cross section shows a decreasing trend, and the chord length of the cross section is less than the preset chord length. The first preset value is less than the second preset value.

6. The route planning method as described in claim 5, characterized in that, When the partition is the first partition, the process of determining the route planning strategy suitable for each partition and performing route planning for the corresponding partition based on the route planning strategy to obtain the route corresponding to each partition includes: Based on the preset safe flight distance, the field of view of the camera on the UAV and the minimum image overlap rate, the effective coverage width of a single image on the blade surface is determined, and the first waypoint step size is determined based on the effective coverage width and the minimum image overlap rate. The three-dimensional curved surface of the wind turbine blade is unfolded into a two-dimensional UV plane along the spanwise principal axis of the wind turbine blade. A uniform grid is generated in the two-dimensional UV plane according to the first waypoint step size, and each grid is projected in reverse onto the three-dimensional curved surface to obtain each surface anchor point. Each surface anchor point is offset outward along the normal vector of its corresponding vertex by the preset safe flight distance to generate waypoints; Based on the waypoints corresponding to each surface anchor point, generate the route corresponding to the first partition.

7. The route planning method as described in claim 5, characterized in that, The curvature includes the maximum principal curvature, and each vertex has a corresponding maximum principal curvature; When the partition is the second partition, the process of determining the appropriate route planning strategy for each partition and performing route planning for the corresponding partition based on the route planning strategy to obtain the route for each partition includes: The second waypoint step size is determined based on the maximum permissible line-of-sight deflection angle and the radius of curvature, wherein the radius of curvature is positively correlated with the second waypoint step size; Extract the vertices associated with the maximum curvature in the second partition and connect them to form a feature ridge; Orthogonally slice along the feature ridge line at the second waypoint step size to obtain surface anchor points, and offset the surface anchor points outward by a preset safety distance along the normal vector of the corresponding vertex to generate waypoints; Based on the waypoints corresponding to each of the surface anchor points, a route corresponding to the second partition is generated.

8. The route planning method as described in claim 5, characterized in that, When the partition is the third partition, the process of determining the appropriate route planning strategy for each partition and performing route planning for the corresponding partition based on the route planning strategy to obtain the route for each partition includes: Extract the extreme points of the spanwise coordinates of the wind turbine blades as the tip sphere centers to establish a local spherical coordinate system; Based on the target ground sampling distance, camera pixel size and focal length, the close-up radius is derived in reverse, and a hemispherical virtual envelope surface with a radius equal to the close-up radius is constructed; The spherical azimuth and polar angle step size are calculated based on the preset image overlap rate, and waypoints are generated on the envelope surface of the hemisphere. Based on each waypoint, generate the route corresponding to the third partition.

9. The route planning method as described in claim 5, characterized in that, The method further includes: The camera's principal optical axis vector at each waypoint in the first partition is controlled to remain antiparallel to the normal vector of the corresponding vertex, and the camera's principal optical axis vector is inversely calculated into the camera gimbal's pitch and yaw angles; and, The principal optical axis vector of the camera at each waypoint in the second partition is controlled to be orthogonal to the normal vector of the corresponding vertex, and the principal optical axis vector of the camera is inversely calculated to obtain the pitch angle and yaw angle of the camera gimbal; and, Control the main optical axis of the camera at each waypoint in the third partition to point towards the center of the leaf tip.

10. The route planning method as described in claim 1, characterized in that, The process of stitching together the flight paths corresponding to each partition to obtain the target flight path for the UAV includes: By concatenating the routes corresponding to each partition, the initial route is obtained; Smoothing is performed at the intersection of the routes in each partition to obtain the transition trajectory corresponding to that intersection; Based on the control frequency of the UAV, the transition trajectory is resampled to obtain transition waypoints, and the initial route is updated using the transition waypoints. The updated flight path is input into the dynamics constraint module, which then filters the updated flight path to obtain the target flight path corresponding to the UAV.

11. The route planning method as described in claim 10, characterized in that, The smoothing process at the intersection of the flight paths in each partition, to obtain the transition trajectory corresponding to that intersection, includes: At the intersection of routes in each partition, a preset number of waypoints are extracted to form a transitional neighborhood; Extract the association information of each waypoint in the transition neighborhood to form the state vector of each waypoint. The association information includes the three-dimensional coordinates of the waypoint, the pitch angle and yaw angle of the camera gimbal associated with the waypoint. The transition trajectory is obtained by fitting the state vector of each waypoint using a cubic B-spline basis function.

12. The route planning method as described in claim 10, characterized in that, The step of inputting the updated flight path into the dynamics constraint module, and then filtering the updated flight path through the dynamics constraint module to obtain the target flight path corresponding to the UAV includes: The updated flight path is input into the dynamics constraint module, which then performs threshold verification on the yaw rate, pitch rate, and UAV flight acceleration and deceleration of the camera gimbal associated with each waypoint in the updated flight path. Waypoints exceeding the maximum permissible angular velocity and acceleration / deceleration limits are removed to obtain compliant waypoints; The target flight path of the UAV is constructed based on the compliant waypoints.

13. A route planning device, characterized in that, The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the route planning method as described in any one of claims 1 to 12.

14. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the route planning method as described in any one of claims 1 to 12.

Images