Method and device for positioning inspection of wind turbine tower drum

Abstract

This application relates to the field of wind turbine generator inspection technology, and discloses a method and device for inspecting and locating wind turbine towers. The method includes: establishing a three-dimensional geometric model of the wind turbine tower and calibrating the physical reference points of the wind turbine tower; acquiring the pose data of a UAV using a quantum inertial measurement unit; collecting point cloud data of the wind turbine tower surface using the UAV's lidar; performing point cloud matching on the pose data based on the point cloud data and the three-dimensional geometric model to obtain a point cloud matching result; acquiring image data containing a first reference point using the UAV's gimbal camera and determining the pixel coordinates of the first reference point in the image data; calibrating the pose data based on the pixel coordinates to obtain a reference point calibration result; and locating the UAV to obtain a UAV positioning result. This application solves the technical problems of insufficient positioning accuracy, large cumulative error, poor environmental adaptability, and low operation and maintenance efficiency in related technologies.

Description

Technical Field

[0001] This application relates to the field of wind turbine generator inspection technology, and more specifically, to a method and device for inspecting and locating wind turbine towers. Background Technology

[0002] Wind turbines operate in complex natural environments for extended periods, making them prone to defects such as tower tilting, weld cracking, and blade surface damage. Regular inspections are essential to ensure operational safety. Current mainstream inspection methods often employ drones equipped with visible light / infrared cameras for manual or semi-automatic flight photography, which suffers from the following technical bottlenecks: First, insufficient positioning accuracy: The surfaces of wind turbine towers and blades lack visual texture features, making traditional Simultaneous Localization and Mapping (SLAM) algorithms prone to failure. Global Positioning System (GPS) signals are susceptible to multipath interference near wind turbines and rely on external signal sources. Second, cumulative error: Traditional Inertial Measurement Units (IMUs) exhibit drift, with accumulated errors reaching meter-level after flying tens of meters, failing to meet sub-centimeter-level defect location requirements. Third, poor environmental adaptability: Strong magnetic fields interfere with electronic compasses, and changes in wind direction cause flight trajectory deviations, further exacerbating positioning errors. Fourth, low efficiency: Related solutions require manual labeling of defect locations afterward, making it impossible to correlate defects with physical coordinates in real-time during flight, resulting in high maintenance costs.

[0003] There is currently no effective solution to the aforementioned technical problems. Summary of the Invention

[0004] This application provides a method and apparatus for inspecting and locating wind turbine towers, which at least solves the technical problem in related technologies that solid-state transformers have fixed control targets and are difficult to achieve optimal performance simultaneously under various mixed load scenarios.

[0005] According to one aspect of the embodiments of this application, a method for inspecting and locating a wind turbine tower is provided, comprising: establishing a three-dimensional geometric model of the wind turbine tower and calibrating physical reference points of the wind turbine tower, wherein the physical reference points include at least a first reference point, the first reference point being located at the top and / or middle of the wind turbine tower; acquiring pose data of the UAV through a quantum inertial measurement unit of a UAV; acquiring point cloud data of the surface of the wind turbine tower through a lidar of the UAV; performing point cloud matching on the pose data based on the point cloud data and the three-dimensional geometric model to obtain a point cloud matching result; acquiring image data containing the first reference point through a gimbal camera of the UAV and determining the pixel coordinates of the first reference point in the image data; performing reference point calibration on the pose data based on the pixel coordinates to obtain a reference point calibration result; and locating the UAV based on the pose data, the point cloud matching result, and the reference point calibration result to obtain a UAV positioning result.

[0006] Optionally, establishing the three-dimensional geometric model of the wind turbine tower includes: establishing the three-dimensional geometric model based on the design parameters of the wind turbine tower and / or the initial flight scan data of the UAV, wherein the wind turbine tower includes a tower body and blades, and the geometric model parameters of the tower body include at least one of the cylindrical axis equation of the tower, the tower radius, and the tower height range; the geometric model parameters of the blades include the three-dimensional surface equation of the blades, and the initial flight scan data includes the coordinate data of the physical reference point, which is obtained by measuring the coordinate data of the positioning unit set on the UAV.

[0007] Optionally, the step of performing point cloud matching on the pose data based on the point cloud data and the three-dimensional geometric model to obtain a point cloud matching result includes: performing point cloud registration on the point cloud data and the three-dimensional geometric model to obtain a point cloud registration result, and determining the first position information of the UAV based on the point cloud registration result; comparing the first position information with the second position information in the pose data to determine the first positioning deviation of the UAV; and if the first positioning deviation is greater than a preset deviation threshold, determining that the quantum inertial measurement unit has a positioning defect.

[0008] Optionally, the step of calibrating the pose data based on the pixel coordinates to obtain the reference point calibration result includes: determining the predicted coordinates of the first reference point based on the pixel coordinates and the pose data of the UAV; comparing the predicted coordinates of the first reference point with the calibrated coordinates of the first reference point to determine the second positioning deviation of the quantum inertial measurement unit.

[0009] Optionally, the method further includes: performing defect detection on the wind turbine tower based on the image data, identifying tower defects, and determining the pixel defect location of the tower defects; acquiring the pose data of the UAV, the ranging data of the lidar, and the gimbal attitude data of the gimbal camera corresponding to the tower defects, wherein the pose data includes the position data and attitude data of the UAV, and the ranging data is the distance from the UAV to the tower surface; mapping the pixel defect location, the pose data, the ranging data, and the gimbal attitude data to the three-dimensional geometric model to determine the physical defect location of the tower defects.

[0010] Optionally, mapping the pixel defect location, the pose data, the ranging data, and the gimbal attitude data to the three-dimensional geometric model to determine the physical defect location of the tower defect includes: determining the optical center position of the gimbal camera, wherein the optical center position is determined based on the pose data and the installation position between the camera optical center and the center of the UAV; determining the direction of light rays pointing from the camera optical center to the tower defect based on the camera optical center position; constructing a ray parameter equation based on the ray direction; determining the tower cylindrical surface equation based on the three-dimensional geometric model; and solving the ray parameter equation and the tower cylindrical surface equation to obtain the three-dimensional physical coordinates of the tower defect.

[0011] Optionally, determining the direction of the light ray pointing from the camera's optical center to the tower defect based on the camera's optical center position includes: converting the pixel coordinates into a normalized direction vector in the camera coordinate system, wherein the normalized direction vector is represented as:

[0012] ;

[0013] In the formula, , Let K be the pixel coordinates of the UAV, and K be the camera intrinsic parameter matrix. Transform the normalized direction vector to the world coordinate system to obtain the physical direction vector of the ray direction, where the physical direction vector is represented as:

[0014] ;

[0015] In the formula, For the drone's attitude, Gimbal posture;

[0016] The step of constructing the ray parameter equation based on the ray direction includes: constructing the ray parameter equation based on the camera optical center position and the physical direction vector, wherein the ray parameter equation is expressed as:

[0017] ;

[0018] In the formula, This is the position of the camera's optical center.

[0019] Optionally, after solving the ray parameter equation and the tower cylindrical surface equation to obtain the three-dimensional physical coordinates of the tower defect, the method further includes: converting the three-dimensional physical coordinates of the tower defect into physical parameters in a cylindrical unfolded coordinate system, wherein the physical parameters include the defect height and the defect angle.

[0020] Optionally, after solving the ray parameter equation and the tower cylindrical surface equation to obtain the three-dimensional physical coordinates of the tower defect, the method further includes: marking the three-dimensional physical coordinates on the three-dimensional geometric model; and / or generating a tower inspection report based on at least one of the physical defect location, defect size, defect type, and defect change trend of the tower defect.

[0021] According to another aspect of the embodiments of this application, an inspection and positioning device for wind turbine towers based on cluster partitioning is also provided, comprising: a model building module configured to establish a three-dimensional geometric model of the wind turbine tower and calibrate the physical reference points of the wind turbine tower, wherein the physical reference points include at least a first reference point, the first reference point being located at the top and / or middle of the wind turbine tower; an acquisition module configured to acquire the pose data of the UAV through the quantum inertial measurement unit of the UAV; and a point cloud acquisition module configured to acquire point cloud data of the surface of the wind turbine tower through the lidar of the UAV. The point cloud matching module is configured to perform point cloud matching on the pose data based on the point cloud data and the three-dimensional geometric model to obtain a point cloud matching result; the pixel recognition module is configured to acquire image data containing the first reference point through the gimbal camera of the UAV and determine the pixel coordinates of the first reference point in the image data; the calibration module is configured to perform reference point calibration on the pose data based on the pixel coordinates to obtain a reference point calibration result; and the positioning module is configured to position the UAV based on the pose data, the point cloud matching result, and the reference point calibration result to obtain a UAV positioning result.

[0022] According to another aspect of the embodiments of this application, a non-volatile storage medium is also provided, the non-volatile storage medium including a stored program, wherein, when the program is running, it controls the device where the non-volatile storage medium is located to execute any of the above-described wind turbine tower inspection and positioning methods.

[0023] According to another aspect of the embodiments of this application, a computer device is also provided, the computer device including a processor, the processor being used to run a program, wherein the program executes any of the above-described methods for inspecting and locating wind turbine towers.

[0024] According to another aspect of the embodiments of this application, a computer program product is also provided, including a computer program that, when executed by a processor, implements any of the above-described methods for inspecting and locating wind turbine towers.

[0025] The wind turbine tower inspection and positioning method and apparatus provided in this application embodiment establishes a three-dimensional geometric model of the wind turbine tower and calibrates the physical reference points of the wind turbine tower. The physical reference points include at least a first reference point, which is located at the top and / or middle of the wind turbine tower. The method acquires the pose data of the UAV using its quantum inertial measurement unit. It also collects point cloud data of the wind turbine tower surface using the UAV's lidar. Based on the point cloud data and the three-dimensional geometric model, it performs point cloud matching on the pose data to obtain a point cloud matching result. Finally, it acquires image data containing the first reference point using the UAV's gimbal camera and determines the pixel coordinates of the first reference point in the image data. The process involves: calibrating the pose data using the pixel coordinates to obtain a reference point calibration result; and positioning the UAV using the pose data, the point cloud matching result, and the reference point calibration result to obtain the UAV positioning result. This achieves high-precision, error-free autonomous positioning in textureless and complex electromagnetic environments, overcoming the limitations of traditional visual SLAM on textureless tower surfaces, eliminating long-term flight accumulation errors of traditional inertial measurement units, resolving GPS multipath interference, and achieving real-time accurate physical coordinate mapping of defects. This also addresses the technical problems of insufficient positioning accuracy, large cumulative errors, poor environmental adaptability, and low maintenance efficiency caused by the inability to correlate defects with physical coordinates in real time. Attached Figure Description

[0026] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings:

[0027] Figure 1 A hardware structure block diagram of a computer terminal for implementing an inspection and positioning method for wind turbine towers is shown.

[0028] Figure 2 This is a flowchart illustrating the inspection and positioning method for wind turbine towers provided in the embodiments of this application.

[0029] Figure 3 This is another flowchart illustrating the inspection and positioning method for wind turbine towers provided in the embodiments of this application;

[0030] Figure 4 This is a structural block diagram of a wind turbine tower inspection and positioning device based on cluster partitioning, provided in an embodiment of this application. Detailed Implementation

[0031] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0032] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0033] According to an embodiment of this application, a method for inspection and positioning of wind turbine towers based on cluster partitioning is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0034] The method embodiment provided in Embodiment 1 of this application can be executed on a mobile terminal, computer terminal, or similar computing device. Figure 1 A hardware block diagram of a computer terminal for implementing an inspection and positioning method for wind turbine towers is shown. Figure 1As shown, the computer terminal 10 may include one or more processors (shown as 102a, 102b, ..., 102n in the figure) (the processor may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, or a camera. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0035] It should be noted that the aforementioned one or more processors or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits can be implemented wholly or partially as software, hardware, firmware, or any other combination. Furthermore, the data processing circuits can be a single, independent processing module, or wholly or partially integrated into any other element in the computer terminal 10. As involved in the embodiments of this application, the data processing circuit serves as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0036] The memory 104 can be used to store software programs and modules of application software, such as the program instructions / data storage device corresponding to the wind turbine tower inspection and positioning method in this embodiment. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby realizing the wind turbine tower inspection and positioning method of the aforementioned application. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor, and these remote memories can be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0037] The display can be, for example, a touchscreen liquid crystal display (LCD) that allows the user to interact with the user interface of the computer terminal 10.

[0038] First, to facilitate understanding of the embodiments of the present invention, some terms or nouns involved in the present invention will be explained below:

[0039] SLAM stands for Simultaneous Localization and Mapping. In the field of SLAM, traditional visual SLAM usually refers to classic methods based on geometry and optimization theory, which is different from the visual SLAM based on deep learning that has emerged in recent years.

[0040] Traditional inertial measurement units (IMUs) typically refer to hardware sensors based on classical physical principles (Newtonian mechanics), which differs from IMU signal processing methods based on deep learning or virtual synthesis.

[0041] An inertial navigation system (INS) is an autonomous system that does not rely on external signals (such as GPS or radio) and does not radiate energy to the outside.

[0042] The Global Positioning System (GPS) is an absolute positioning system based on satellite radio ranging. In navigation systems, as a global absolute position observation sensor, its core function is to provide absolute spatial reference information based on the Global Navigation Satellite System (GNSS).

[0043] An inertial navigation system (INS) is a fully autonomous navigation system. Its core, the inertial measurement unit (IMU), has the advantages of requiring no external signals and being resistant to interference. However, because errors accumulate and drift over integration time, it is usually combined with GNSS (such as GPS) in engineering applications to ensure positioning accuracy and reliability.

[0044] Figure 2 and Figure 3 This is a flowchart illustrating the inspection and positioning method for wind turbine towers provided in the embodiments of this application, as shown below. Figure 2 and Figure 3 As shown, the method includes the following steps:

[0045] Step S201: Establish a three-dimensional geometric model of the wind turbine tower and mark the physical reference points of the wind turbine tower. The physical reference points include at least a first reference point, which is set at the top and / or middle of the wind turbine tower.

[0046] In the embodiments of this application, unless otherwise stated, the term "wind turbine tower" generally refers to the main supporting structure of a wind turbine, including the tower body and the blades attached thereto.

[0047] In the preparation phase before the UAV inspection flight, a three-dimensional geometric model of the wind turbine tower is first established. This model can be based on the wind turbine's factory design parameters (such as cylinder radius, height, and axis coordinates) or reconstructed from the point cloud scanned during the initial UAV inspection. Simultaneously, the physical reference points of the wind turbine tower are calibrated. Specifically, a high-precision positioning unit (such as RTK-GPS or a total station) is used to determine the absolute three-dimensional coordinates of preset markers in the tower coordinate system, and these coordinates are stored in the system database. Preferably, the first reference point is set at the top of the tower (such as at the nacelle base) or the middle (such as at the flange connection) to provide a high-altitude attitude calibration reference during flight.

[0048] It should be noted that physical reference points can be high-contrast visual markers posted manually, or natural geometric feature points on the tower structure (such as the center of flange bolt holes), as long as their spatial position can be measured with high precision.

[0049] In this embodiment, the high-precision positioning unit can be a Real-Time Kinematic Global Positioning System (RTK-GPS), a total station, or a combination of both. It is used during the offline preparation phase to accurately measure the three-dimensional absolute coordinates of preset physical reference points (such as QR codes) at the bottom, middle, and top of the tower, as well as the position of the tower's central axis. This high-precision data is used for subsequent calibration of the quantum inertial navigation system's initial null position and online error correction, and provides a comparison standard for subsequent point cloud matching and reference point calibration.

[0050] RTK-GPS is a real-time dynamic positioning technology based on carrier phase observations. It acquires satellite signals through a base station receiver, calculates corrections, and transmits them in real-time to a mobile station (drone). The mobile station uses these corrections to differentially process its own GPS observations, thereby eliminating common errors such as ionospheric delay, tropospheric delay, and satellite orbital errors, achieving high-precision positioning at the centimeter or even millimeter level.

[0051] A total station is a high-tech surveying instrument integrating optics, mechanics, and electronics. It combines the functions of an electronic distance meter and an electronic theodolite, capable of simultaneously measuring horizontal angles, vertical angles, and slope distances. Its built-in microprocessor computer calculates and displays data such as distance, elevation difference, and coordinates. Its core advantage lies in its extremely high accuracy in angle and distance measurement (typically at the millimeter level), making it commonly used to establish high-precision control networks or calibrate the absolute coordinates of key points. Compared to RTK-GPS, total stations offer higher measurement accuracy and stability within a small, localized area, making them particularly suitable as a reference source for high-precision indoor or near-field calibration, providing reliable initial zero points and calibration true values ​​for quantum inertial navigation systems.

[0052] Step S202: Acquire the pose data of the UAV through the quantum inertial measurement unit of the UAV.

[0053] Before a UAV performs an inspection mission, initial calibration must be completed to establish a global absolute coordinate reference. Specifically, a second physical reference point (e.g., a highly reflective QR code) is preset at the bottom of the tower (such as the edge of a concrete foundation), and the precise calibration coordinates of this reference point in the global world coordinate system are obtained using a high-precision total station or RTK-GPS. This second reference point serves as the starting anchor point for the positioning process, providing an initial absolute position reference relative to the first physical reference point located in the middle and top of the tower. Before takeoff, the UAV must be moved to the second physical reference point (touching or hovering at a known height directly above it), and the initial position state vector of the Quantum Inertial Measurement Unit (Q-IMU) is directly set to the precise calibration coordinates of this second physical reference point through the navigation system. At the same time, the initial attitude angles (such as pitch and roll zeroing, and yaw aligned with the reference direction) are set in conjunction with the tower's vertical or horizontal information. This initialization process sets the initial state (position, attitude, velocity) of the navigation system to known true values, establishes a strict alignment relationship between the navigation coordinate system and the global world coordinate system, effectively eliminates the initial alignment error and position deviation of quantum inertial navigation in the startup phase, and provides a high-precision absolute position reference for subsequent drift-free autonomous navigation based on quantum inertial navigation.

[0054] During UAV inspection flights, if the UAV is flying along a planned path of spiraling upwards around the tower, the onboard quantum inertial measurement unit (Q-IMU) acts as the core navigation sensor, collecting the UAV's three-axis angular velocity and three-axis acceleration data in real time at a fixed frequency (e.g., 200Hz). The Q-IMU utilizes quantum mechanical effects such as atomic interference and atomic spin to sense inertial motion. A typical cold atom Q-IMU manipulates atomic quantum states using laser pulses and measures interference phase changes, thereby achieving high-precision sensing of the carrier's motion. The Q-IMU uses an internal integration algorithm to calculate the UAV's position relative to the initial coordinate system in real time. and posture (i.e., pose data).

[0055] From a system function perspective, Q-IMU is a new type of IMU, which includes an atomic interferometer gyroscope and a quantum accelerometer. Its positioning error is extremely small (e.g., less than 1 meter / 100 hours). Furthermore, by combining Q-IMU with inertial navigation calculation, multi-source data fusion, etc., a quantum-enhanced inertial navigation system can be formed, which will be referred to as quantum inertial navigation or inertial navigation for short.

[0056] Compared to traditional inertial measurement units (IMUs), Q-IMUs exhibit extremely low zero-bias instability and extremely high long-term stability, solving the problem of meter-level positioning errors caused by error accumulation (drift) over time in traditional IMUs. In wind turbine tower and blade inspection scenarios, facing challenges such as GPS signal interference from multipath effects or the lack of visual texture features on the tower surface, Q-IMUs can provide continuous, stable, and drift-free pose data in environments without external signal (GPS) assistance. This high-precision positioning not only ensures the autonomous navigation capability of UAVs in complex environments but also provides high-precision reference pose data for subsequent LiDAR point cloud matching, reference point visual calibration, and real-time physical coordinate mapping of defects.

[0057] Preferably, to further optimize the multi-sensor fusion effect, during the initialization phase, the lidar can be controlled to scan the bottom reference point, and the initial matching relationship between the lidar coordinate system and the world coordinate system can be established by utilizing the geometric features of the known reference point.

[0058] Step S203: Collect point cloud data of the wind turbine tower surface using the lidar of the drone.

[0059] During the UAV inspection flight, the onboard lidar sensor emits laser pulses into the surrounding space and receives the echo signals reflected from the surface of the wind turbine tower. By processing the time difference between transmission and reception and the signal strength, the system acquires three-dimensional point cloud data of the tower surface. This point cloud data consists of a large number of points with spatial coordinates. The points formed by these points accurately characterize the actual geometry and surface details of the tower surface, providing raw observation data for subsequent environmental perception and positioning calibration.

[0060] In this embodiment, the point cloud data is acquired in two ways, including but not limited to: one is to directly acquire dense three-dimensional point clouds using 3D LiDAR; the other is to use a single-line LiDAR to scan the trajectory multiple times using a UAV, and then stitch together and reconstruct a three-dimensional point cloud from the collected two-dimensional cross-sectional data. Regardless of the method used, the final point cloud data contains the spatial geometric information of the tower surface, which is used to compare with the preset three-dimensional geometric model of the tower to verify the accuracy of the position calculated by the quantum inertial measurement unit and to provide necessary geometric constraints to correct its long-term accumulated errors.

[0061] Step S204: Perform point cloud matching on the pose data based on point cloud data and 3D geometric model to obtain point cloud matching results.

[0062] In this embodiment, the point cloud data of the tower surface collected in real time by the lidar is combined with the pose data provided by the UAV's quantum inertial measurement unit (Q-IMU) at the current moment and transformed into a unified world coordinate system through coordinate transformation. Then, a point cloud registration algorithm or a feature matching-based algorithm is used to match the transformed actual point cloud data with the pre-stored three-dimensional geometric model of the tower.

[0063] The algorithm iteratively optimizes the Euclidean distance error between the actual point cloud and the model surface to calculate the optimal spatial pose transformation relationship. The resulting point cloud matching result is represented as a relative pose transformation matrix or position deviation vector. This result reflects the spatial alignment between the observed tower geometry and the theoretical model, and further derives the precise lateral position and attitude deviation of the UAV relative to the tower axis.

[0064] This step utilizes the inherent cylindrical geometry of the tower to provide independent observations based on environmental structural information. These observations do not depend on external signals (such as GPS) or visual texture features, and can effectively verify and constrain the lateral drift of the quantum inertial navigation-derived position in the direction perpendicular to the tower axis. This is a core step in geometric model-assisted verification and improves the robustness of UAV positioning in complex environments.

[0065] Step S205: Acquire image data containing the first reference point using the gimbal camera of the drone, and determine the pixel coordinates of the first reference point in the image data.

[0066] In this embodiment, when the UAV flies to a preset height (i.e., the first reference point) at the top or middle of the tower according to the planned path, the gimbal camera is controlled to capture image data containing the first reference point. The two-dimensional pixel coordinates of the first reference point in the image are extracted using an image recognition algorithm (such as QR code recognition or feature point detection). These pixel coordinates reflect the relative position information of the first reference point in the current camera field of view.

[0067] Determining pixel coordinates includes sub-steps such as image preprocessing, feature extraction, and coordinate transformation.

[0068] This step utilizes visual information to provide a high-precision absolute position reference. Since the precise calibration coordinates of the first physical reference point in three-dimensional space (pre-measured and stored in a database by RTK-GPS or a total station) are known, the theoretical projected position of the reference point can be calculated by combining the acquired pixel coordinates with camera intrinsic parameters and the UAV's real-time pose. By comparing the theoretical projected position with the actual detected pixel coordinates, or by using the principle of spatial resection to solve for the UAV's position, the absolute position deviation between the UAV's current estimated position and its true position can be calculated. This deviation serves as a key observation value, used to subsequently correct the accumulated errors generated by quantum inertial navigation, ensuring the global consistency of the navigation system.

[0069] Step S206: Perform reference point calibration on the pose data based on pixel coordinates to obtain the reference point calibration result.

[0070] In this embodiment, after obtaining the pixel coordinates of the first reference point in the image, it is combined with the real-time pose data of the UAV (including position) provided by the quantum inertial measurement unit (Q-IMU) at the current moment. and posture The system uses a camera imaging geometry model to inversely calculate the reference point's coordinates in the UAV's body coordinate system or world coordinate system. Then, it compares the calculated reference point coordinates with pre-stored, precisely calibrated coordinates obtained from RTK-GPS or a total station, calculating the position deviation vector between the two. This deviation vector is the reference point calibration result, reflecting the cumulative error or drift generated by the quantum inertial navigation system after long-term flight, and is used to correct navigation parameters online.

[0071] This step provides direct, high-precision absolute coordinate constraints through visual reference points, solving the problems of high-frequency noise in quantum inertial navigation (QINa) despite its lack of long-term drift, and the lack of an absolute reference frame in lidar point cloud matching, which can constrain relative geometry. Furthermore, the visual information is converted into state correction quantities usable by the navigation system for online correction of the QINa's navigation parameters, ensuring global consistency and absolute accuracy of the UAV's positioning results.

[0072] Step S207: Based on pose data, point cloud matching results, and reference point calibration results, the UAV is located to obtain the UAV positioning result.

[0073] To obtain high-precision final positioning results, a multi-source data fusion approach is adopted. The pose data in step S202, the point cloud matching results in step S204, and the reference point calibration results in step S206 are fused using the Extended Kalman Filter (EKF) framework to generate the final high-precision UAV positioning results.

[0074] Among them, the Extended Kalman Filter (EKF) is a recursive estimation algorithm for nonlinear systems, suitable for scenarios like UAVs where the equations of motion are nonlinear. Fusion refers to the weighted combination of information from multiple sources, with the weights typically determined by the uncertainties (covariance) of the sensors.

[0075] Specifically, the pose data calculated by the quantum inertial measurement unit (Q-IMU) is used as the state prediction value of the filter, leveraging its high-frequency output and high confidence characteristics to provide a continuous pose trajectory. Simultaneously, the relative position observations obtained from point cloud matching and the absolute position observations obtained from reference point calibration are used as measurement update terms for the filter. During the filtering process, the optimal solution for state estimation is calculated in real time, thereby outputting the fused high-precision position and attitude of the UAV in the world coordinate system.

[0076] This step fully leverages the advantages of each sensor through a multi-source fusion mechanism: on the one hand, the high-frequency continuity of quantum inertial navigation ensures the real-time performance and smoothness of positioning; on the other hand, by matching the lidar point cloud with the geometric model, the lateral drift of the UAV in environments lacking texture features is effectively constrained, preventing positioning divergence; in addition, by using the absolute position calibration of physical reference points, the cumulative errors that quantum inertial navigation may generate during long-term flight are eliminated.

[0077] In an optional embodiment, step S201, establishing a three-dimensional geometric model of the wind turbine tower, includes: establishing a three-dimensional geometric model based on the design parameters of the wind turbine tower and / or the initial flight scan data of the UAV, wherein the wind turbine tower includes the tower body and blades, and the geometric model parameters of the tower body include at least one of the cylindrical axis equation of the tower, the tower radius, and the tower height range; the geometric model parameters of the blades include the three-dimensional surface equation of the blades; the initial flight scan data includes the coordinate data of physical reference points, which are obtained by measuring the coordinate data of the positioning unit set on the UAV.

[0078] In this embodiment, the three-dimensional geometric model of the wind turbine tower can be constructed using one of the following two methods or a combination thereof:

[0079] (1) Modeling based on the design parameters of the wind turbine tower. Directly obtain the tower design parameters and blade design data provided by the wind turbine manufacturer. The design parameters may include CAD model files, dimensions in engineering drawings, tower cross-sectional radius variation curves, blade airfoil data and three-dimensional surface equations, etc.

[0080] (2) Modeling based on initial flight scan data from UAVs. Before the formal inspection mission, a calibration / database building mission is performed. The UAV carries a high-precision positioning unit to perform an all-round scan of the tower and blades, collecting multi-view images or high-density point cloud data. A high-precision three-dimensional geometric model is generated by point cloud reconstruction algorithms (such as Poisson reconstruction, Delaunay triangulation, etc.), and geometric feature parameters are extracted from it.

[0081] The three-dimensional geometric model of the wind turbine tower encompasses both the tower body and the blades. Details are as follows:

[0082] The geometric model parameters of the tower body are used to describe the spatial geometric characteristics of the tower. Specifically, these parameters may include the equation of the cylindrical axis of the tower's central axis (in three-dimensional space, a straight line can be represented by a point plus a direction vector, or by two points), used to determine the spatial attitude and position of the tower; the tower radius, used to define the radial distance of the tower surface; and the tower height range, used to define the effective inspection range of the tower. In practical applications, these parameters are usually included simultaneously to construct a complete cylindrical model.

[0083] The geometric model parameters of the blade describe the spatial geometry of the blade surface, mainly including the three-dimensional surface equations of the blade (which can be NURBS surfaces, B-Spline surfaces, or parametric surface equations). These equations are used for point cloud matching on the blade surface and for locating defects in the blade's local coordinate system.

[0084] In this embodiment, the initial flight scan data also includes the coordinate data of physical reference points. This coordinate data is obtained through measurements taken by the high-precision positioning unit onboard the UAV during the initial database construction flight, and serves as the absolute reference source for visual recognition and position calibration during subsequent inspections. Preferably, when the UAV is equipped with an RTK (Real-Time Kinematic) module, reference point coordinates with centimeter-level accuracy can be obtained.

[0085] The aforementioned high-precision positioning unit can be at least one of the Global Navigation Satellite System (GNSS) receivers (such as GPS, BeiDou, etc.), or a combination module of inertial navigation system and GNSS.

[0086] In an optional embodiment, step S204 involves performing point cloud matching on the pose data of the quantum inertial measurement unit based on point cloud data and a three-dimensional geometric model to obtain point cloud matching results, including:

[0087] S2041, perform point cloud registration with the 3D geometric model to obtain the point cloud registration result, and determine the first position information of the UAV based on the point cloud registration result;

[0088] S2042, compare the first position information with the second position information in the pose data to determine the first positioning deviation of the UAV;

[0089] S2043, if the first positioning deviation is greater than the preset deviation threshold, it is determined that the quantum inertial measurement unit has a positioning defect.

[0090] In step S2041, the currently acquired point cloud data is spatially aligned with the pre-stored 3D geometric model of the tower using a point cloud registration algorithm, such as the Iterative Closest Point (ICP) algorithm, to generate a point cloud registration result. This registration result reflects the spatial transformation relationship between the point cloud data and the model. Based on this transformation relationship, the spatial position of the UAV in the tower coordinate system is calculated, i.e., the UAV's first position information.

[0091] Point cloud registration refers to the process of aligning two sets of point cloud data or point cloud data with model data, typically achieved by minimizing the distance error between the two point sets. The Iterative Closest Point (ICP) algorithm is a fine point cloud registration method that iteratively optimizes the rotation matrix. With translation vector The algorithm minimizes the distance between corresponding points to achieve optimal overlap between the source and target point clouds. It uses the coarse registration output as the initial value for iteration and gradually converges to fine registration. In this application, the algorithm can provide a precise spatial reference for navigation and positioning. The first position information refers to the UAV position indirectly calculated through geometric matching, which differs from the position obtained directly through inertial navigation integration.

[0092] In step S2042, the second position information (i.e., the UAV position calculated by the inertial navigation system) is extracted from the pose data output by the quantum inertial measurement unit. The first position information obtained based on point cloud registration is compared with the second position information by spatial distance calculation or vector comparison to obtain the difference between the two, i.e., the first positioning deviation.

[0093] The comparison can be done using either Euclidean distance calculation or attitude angle difference calculation. The first positioning deviation refers to the error between the geometrically observed position and the inertial navigation system-calculated position.

[0094] This step is used to determine whether the quantum inertial navigation system is functioning correctly or requires calibration. It's understandable that a larger initial positioning deviation indicates less accurate inertial navigation calculations.

[0095] In step S2043, the calculated first positioning deviation is compared with a preset deviation threshold. If the first positioning deviation is greater than the preset deviation threshold, the current positioning accuracy of the quantum inertial measurement unit is determined to be unreliable, i.e., there is a positioning defect. At this time, the abnormal state can be recorded, or the subsequent calibration process can be triggered.

[0096] Among them, the preset deviation threshold can be dynamically adjusted or fixedly set according to flight altitude, speed, and environmental complexity, such as 5cm; positioning defects refer to the state of reduced reliability of inertial navigation output data, caused by factors such as hardware failure, extreme environmental interference, or long-term drift.

[0097] In an optional embodiment, step S206 involves calibrating the pose data based on pixel coordinates to obtain a reference point calibration result, including:

[0098] S2061, Determine the predicted coordinates of the first reference point based on pixel coordinates and the pose data of the UAV;

[0099] S2062, compare the predicted coordinates of the first reference point with the calibrated coordinates of the first reference point to determine the second positioning deviation of the quantum inertial measurement unit.

[0100] In step S2061, based on the pixel coordinates of the first reference point in the image and combined with the camera intrinsic parameter matrix, the coordinates are converted into a normalized direction vector in the camera coordinate system. Then, combined with the UAV's pose data (including the UAV's position and attitude rotation matrix in the world coordinate system, and the camera's extrinsic parameters relative to the UAV), this direction vector is transformed to the world coordinate system. Using ray casting or triangulation principles, and combined with the known physical properties or spatial constraints of the first reference point, the predicted coordinates of the first reference point in the system coordinate system are calculated. These predicted coordinates reflect the spatial position of the reference point calculated based on the current visual observation and inertial navigation pose.

[0101] It's important to clarify that the predicted coordinates here do not refer to a temporal prediction, but rather to the ideal position of the reference point calculated based on current sensor data (vision + inertial navigation). In other words, these coordinates are calculated based on pose data with potential errors, therefore they deviate from the true calibration coordinates and are used for subsequent calibration. Understandably, if the inertial navigation pose is accurate, the predicted coordinates should be very close to the true physical coordinates of the reference point.

[0102] In step S2062, the predicted coordinates of the first reference point, calculated based on visual observation, are compared with the pre-stored precise calibration coordinates of the first reference point using spatial distance calculation or vector comparison to obtain the difference between the two, i.e., the second positioning deviation. The second positioning deviation reflects the positioning error of the quantum inertial measurement unit at the current moment and can be used to correct the pose data of the quantum inertial measurement unit online or as an observation value for extended Kalman filtering.

[0103] It should be noted that the second positioning deviation refers to the error observation value introduced by the calibration of the visual reference point. Since the coordinates of the reference point are known with high precision, unlike the first positioning deviation (introduced by point cloud matching), this deviation focuses more on the calibration of the absolute position.

[0104] This step is a concretization of the benchmark calibration results. The first positioning deviation obtained through point cloud matching and the second positioning deviation obtained through benchmark visual calibration will be fused to optimize the final positioning result.

[0105] In one optional embodiment, the inspection and positioning method for wind turbine towers further includes:

[0106] S301, based on image data, performs defect detection on the wind turbine tower, identifies tower defects, and determines the pixel defect location of the tower defects;

[0107] S302 acquires the pose data of the UAV, the ranging data of the LiDAR, and the gimbal attitude data of the gimbal camera corresponding to the tower defect. The pose data includes the position data and attitude data of the UAV, and the ranging data is the distance from the UAV to the tower surface.

[0108] S303 maps the pixel defect location, pose data, ranging data, and gimbal attitude data to a three-dimensional geometric model to determine the physical defect location of the tower defect.

[0109] In step S301, an optical / infrared camera is used to acquire images in real time, and computer vision algorithms (such as semantic segmentation networks based on deep learning or traditional feature matching algorithms) are used to process the image data acquired in step S201 to identify defect areas (such as cracks, corrosion, and spalling) on ​​the surface of the wind turbine tower. For each identified defect, its two-dimensional pixel coordinates in the image coordinate system are calculated, i.e., the pixel defect location. This position represents the index of the defect in the image matrix.

[0110] This step identifies the pixel-level defect location of the tower defect, which is the key and objective of UAV inspection.

[0111] In step S302, while identifying tower defects, the following sensor data is recorded simultaneously: the UAV's pose data, including the UAV's three-dimensional position in the world coordinate system at the current moment. and attitude rotation matrix The vertical distance from the UAV to the tower surface at the current moment, as measured by lidar. (Or the average distance to the corresponding defect area in the point cloud); real-time attitude data of the gimbal camera relative to the UAV body (such as pitch angle, yaw angle, roll angle, or rotation matrix). These data have the same timestamp, ensuring the accuracy of spatial geometric relationships.

[0112] This step provides the spatiotemporal reference frame data and geometric constraint data required for defect localization.

[0113] In step S303, the pixel defect position is converted into a direction vector in the camera coordinate system using the camera imaging geometric model. Combining the gimbal attitude data and UAV pose data, the direction vector is converted to the world coordinate system, and the ray equation pointing from the camera optical center to the defect is constructed. At the same time, combined with the lidar ranging data (used to determine the intersection distance between the ray and the tower surface or for verification), the ray is intersected with the three-dimensional geometric model of the tower (such as the cylindrical surface equation) to obtain the three-dimensional physical coordinates of the defect on the tower surface, i.e., the physical defect position.

[0114] Mapping to a three-dimensional geometric model refers to a continuous transformation of the coordinate system. The location of the physical defect includes both rectangular coordinates and cylindrical unfolded coordinates.

[0115] This step transforms visual defects into physical coordinates that are understandable to operations and maintenance personnel.

[0116] In one optional embodiment, the pixel defect location, pose data, ranging data, and gimbal attitude data are mapped onto a three-dimensional geometric model to determine the physical defect location of the tower defect, including:

[0117] S401, Determine the optical center position of the gimbal camera, whereby the optical center position is determined based on pose data and the installation position between the optical center of the camera and the center of the drone;

[0118] S402, determine the direction of the light rays pointing from the camera's optical center to the tower defect based on the camera's optical center position;

[0119] S403, constructing ray parameter equations based on ray direction;

[0120] S404, Determining the equation of the tower's cylindrical surface based on a three-dimensional geometric model;

[0121] S405, solve the ray parameter equation and the tower cylindrical surface equation to obtain the three-dimensional physical coordinates of the tower defect.

[0122] In step S401, based on the UAV's pose data, and combined with the pre-calibrated installation translation vector of the camera's optical center relative to the UAV's center (or center of gravity), Through coordinate transformation formula Calculate the absolute position of the camera's optical center in the world coordinate system. This step establishes the spatial relationship between the world coordinate system and the camera's optical center, which is used for subsequent visual geometry calculations.

[0123] The camera optical center is the optical center of the camera lens and the theoretical starting point for light to enter the camera. The installation position refers to the offset of the camera coordinate system origin from the UAV coordinate system, and is part of the camera extrinsic parameter calibration.

[0124] In step S402, the pixel coordinates of the defect are converted into a normalized direction vector in the camera coordinate system using the camera intrinsic parameter matrix. Then, combined with the gimbal attitude data and UAV attitude data, the direction vector is rotated to the world coordinate system to obtain the ray direction vector pointing from the camera optical center to the tower defect.

[0125] In three-dimensional space, the direction of a ray of light is represented by a vector that describes the path of light propagation. Determining the direction of a ray of light requires combining gimbal attitude data and UAV attitude data.

[0126] In an optional embodiment, in S402, determining the direction of the light ray pointing from the camera's optical center to the tower defect based on the camera's optical center position includes:

[0127] S4021 converts pixel coordinates into normalized direction vectors in the camera coordinate system, where the normalized direction vector is expressed as:

[0128] ;

[0129] In the formula, , The pixel coordinates of the drone. This is the camera intrinsic parameter matrix;

[0130] S4022, transform the normalized direction vector to the world coordinate system to obtain the physical direction vector of the ray direction, where the physical direction vector is expressed as:

[0131] ;

[0132] In the formula, For the drone's attitude, Gimbal posture;

[0133] In S4021, a normalized direction vector with a magnitude of 1 represents only the direction. (Internal parameter matrix) It is a matrix that describes the internal optical characteristics of a camera, and is usually obtained from camera calibration. It is a homogeneous representation of pixel coordinates on the image. This step establishes a mathematical connection between image pixels and camera spatial geometry, transforming visual information into spatial geometric information.

[0134] In S4022, attitude is specifically represented by a rotation matrix, describing the relative orientation of the coordinate system. (UAV attitude rotation matrix) Used to represent the transformation from the UAV's body coordinate system to the world coordinate system. Gimbal attitude rotation matrix. This is used to represent the transformation from the camera coordinate system to the UAV body coordinate system, or the relative attitude of the gimbal to the UAV. This step fully considers the influence of the UAV's flight attitude and gimbal rotation on the camera's pointing, and completes the transformation from the local sensor coordinate system to the global world coordinate system, ensuring the correctness of the light direction in the unified reference system (world system).

[0135] In step S403, the parametric equations of the light rays are constructed, starting from the camera's optical center position and taking the ray direction vector as the direction. The parametric equations of light rays are the standard mathematical form describing straight lines in three-dimensional space, transforming geometric light rays into algebraic equations for easier solving.

[0136] In an optional embodiment, in S403, constructing the ray parameter equation based on the ray direction includes: constructing the ray parameter equation based on the camera optical center position and the physical direction vector, wherein the ray parameter equation is expressed as:

[0137] ;

[0138] In the formula, This is the position of the camera's optical center.

[0139] In this embodiment, the camera optical center position As the starting point, the direction vector of light As a direction, Let be a parameter representing the distance along the direction of the light ray. The above formula uses this parameter. This step represents the equation representing the coordinates of any point on the line. It determines the equation of the line used for geometrically solving for the intersection point.

[0140] In step S404, the geometric parameters of the tower body (such as the coordinates of the center axis, radius, and height range) are extracted from the pre-stored three-dimensional geometric model of the tower, and the equation of the cylindrical surface describing the outer surface of the tower is constructed:

[0141] ;

[0142] in, , The coordinates of the tower's central axis are... Where is the tower radius, , This refers to the range of tower heights. These parameters are provided by the geometry model module.

[0143] In step S405, the ray parameter equation is substituted into the tower cylindrical surface equation to construct a parameter... quadratic equation

[0144] ;

[0145] Solve this quadratic equation, choosing the option that satisfies the physical constraints ( and Coordinates at Solution (within the range of tower height, i.e., the intersection point is within the range of tower height). .Will Substituting back into the ray parameter equation, the three-dimensional physical coordinates of the tower defect are calculated. .

[0146] In an optional embodiment, after solving the ray parameter equation and the tower cylindrical surface equation in step S405 to obtain the three-dimensional physical coordinates of the tower defect, the method further includes: converting the three-dimensional physical coordinates of the tower defect into physical parameters in the cylindrical unfolded coordinate system, wherein the physical parameters include the defect height and the defect angle.

[0147] In this embodiment, the obtained three-dimensional physical coordinates of the tower defect are... Subsequently, to facilitate understanding and recording by maintenance personnel, the three-dimensional coordinates were converted into height and circumferential angles in a cylindrical unfolded coordinate system.

[0148] Based on the geometric center of the tower's cylindrical surface and radius This converts the three-dimensional physical coordinates into physical parameters in a cylindrical unfolded coordinate system. Among these parameters is the defect height. Defects can be directly identified. Axis coordinates (or heights relative to a reference plane) are... Defect angle Calculated using the arctangent function: , which represents the azimuth angle of the defect on the circumference of the tower.

[0149] The cylindrical unfolded coordinate system is a two-dimensional coordinate system formed by unfolding the surface of a cylinder into a plane, with the horizontal axis representing angle (or arc length) and the vertical axis representing height. This step transforms technical data into operational data.

[0150] In one optional embodiment, after solving the ray parameter equation and the tower cylindrical surface equation to obtain the three-dimensional physical coordinates of the tower defect, the method further includes:

[0151] S501 marks the three-dimensional physical coordinates on the three-dimensional geometric model.

[0152] In S501, the three-dimensional physical coordinates of the tower defect are obtained, the three-dimensional geometric model display module is called, and a visual marker (such as a red dot, icon or highlighted area) is generated at the coordinate location. The marker is then overlaid and displayed on the three-dimensional geometric model interface of the wind turbine tower to achieve an intuitive display of the defect location.

[0153] Annotation includes any means of marking locations on the model, such as color coding, symbol addition, and 3D entity attachment. A 3D geometric model refers to a digital tower model used for display, typically in formats such as STL, OBJ, and PLY, or objects within the engine.

[0154] This step visualizes the defects to assist in operational decision-making.

[0155] In one optional embodiment, after solving the ray parameter equation and the tower cylindrical surface equation to obtain the three-dimensional physical coordinates of the tower defect, the method further includes:

[0156] S502, generate a tower inspection report based on at least one of the following: physical defect location, defect size, defect type, and defect change trend.

[0157] In S502, the physical location (e.g., height and angle), dimensions (e.g., length and width), and type (e.g., cracks, corrosion) of tower defects are summarized, along with the defect change trend obtained by comparing historical data. Based on at least one of these pieces of information, a structured tower inspection report is automatically generated. The report format may include PDF, HTML, or database records, and is submitted to maintenance personnel for further processing. This step automates the archiving and delivery of inspection data, transforming technical data into document assets within the business process.

[0158] The following specific embodiments illustrate the inspection and positioning method for wind turbine towers provided in this application. Specifically, an onshore wind farm will be used as an example for detailed explanation.

[0159] Step S1, configure the gimbal and camera.

[0160] The drone is equipped with a three-axis stabilized gimbal and a 20-megapixel visible light camera with a 24mm focal length (35mm equivalent). The gimbal has an attitude accuracy of 0.02° and can output the rotation matrix relative to the drone body in real time. The lidar ranging accuracy is ±2cm, with a maximum ranging distance (i.e., point cloud matching result) of 50m. The translation vector of the camera's optical center relative to the UAV's center of gravity. The factory specifications are (0.1m, 0.0m, 0.3m).

[0161] Step S2: Establishment of logistics reference points.

[0162] High-reflectivity QR codes were placed at the edge of the concrete foundation at the bottom of the wind turbine tower (0.5m high), and at the middle flange (30m high) or the top nacelle base (80m high). Their absolute coordinates were measured using RTK-GPS (2cm accuracy) and stored in a database. Simultaneously, the center coordinates of the tower's bottom and top were measured to calculate the tower's axis equation and average radius (measured radius 2.5m).

[0163] Step S3: Quantum inertial navigation system selection.

[0164] Employing the Q-CTRL Ironstone Opal quantum navigation system (a commercial quantum navigation system launched by the Australian quantum technology company Q-CTRL, already integrated into commercial drone platforms), the system is smaller than a $5 bill, and flight tests have verified that its accuracy is 11 times higher than that of traditional inertial navigation systems (INS). The system outputs a frequency of 200Hz, with an attitude angle accuracy of 0.01° and a position accuracy of 0.02m / 100m.

[0165] Step S4, flight path planning.

[0166] The drone takes off from the bottom reference point and flies around the tower along a spiral ascent path at an ascent speed of 0.5 m / s and an angular velocity of 5° / s, ensuring that each square meter of the tower surface is covered by at least 3 images. Every 5 meters of ascent, it pauses for 2 seconds to acquire high-resolution images and complete a lidar scan.

[0167] Step S5: Online calibration.

[0168] When flying over the central flange reference point (30m altitude), the camera recognizes the QR code, and the quantum inertial navigation system calculates the position with a deviation of 1.2cm from the actual coordinates, triggering the Kalman filter for online correction. The UAV's position and attitude data, maximum range measurement, and position-to-actual coordinate deviation data are fused and processed to generate a high-precision positioning result for the UAV, ensuring that the subsequent flight error remains within 2cm.

[0169] Step S6, Defect Detection and Location.

[0170] At an altitude of 42.5m, a 0.5mm wide crack was identified near the blade root. Simultaneous recording: UAV location. UAV attitude (Yaw 30°, Pitch -2°, Roll 1°); Gimbal attitude (Relative to the drone: Yaw 0°, Pitch -15°, Roll 0°); Pixel coordinates Laser ranging .

[0171] First, calculate the position of the camera's optical center:

[0172] ;

[0173] Convert pixel coordinates to direction vectors in the camera coordinate system:

[0174] ;

[0175] Calculations show that (After normalization).

[0176] Then, the above direction vector Transform to world coordinate system:

[0177] ;

[0178] After that Substitute into the light ray parameter equation And the equation of the cylindrical surface of the tower. Solving the system of equations simultaneously, we get... The three-dimensional coordinates of the defect are then:

[0179] ;

[0180] Convert the 3D coordinates of the defect to the cylindrical unfolded coordinates:

[0181] ;

[0182] .

[0183] The above-mentioned cylindrical unfolded coordinates can be used as the output of the tower inspection report. In the generated tower inspection report, the specific information of the defect is output as: "Crack at the blade root, height 42.5m, azimuth angle 38.5°, length 0.5mm".

[0184] Combining the above optional implementation methods achieves at least the following effects:

[0185] (1) Two-level calibration architecture of quantum inertial navigation and physical reference point: Quantum inertial navigation technology is introduced into wind turbine inspection and positioning, and physical reference points are used to eliminate initial errors and periodic calibration, thus realizing drift-free autonomous navigation.

[0186] (2) Geometric model-assisted verification in featureless environment: In view of the textureless characteristics of the tower, the point cloud of the lidar is matched with the pre-stored geometric model as an independent observation source of the quantum inertial navigation, which ensures the positioning reliability.

[0187] (3) Real-time defect location method based on cylindrical surface geometric model: Taking full account of the cylindrical structure characteristics of the tower, the UAV attitude, gimbal pointing, laser ranging and cylindrical surface equation are combined. The precise position (height + circumferential angle) of the defect on the tower surface is calculated by rigorous coordinate transformation formula, realizing "location as soon as it is captured".

[0188] According to an embodiment of this application, a wind turbine tower inspection and positioning device based on cluster partitioning is also provided for implementing the above-described wind turbine tower inspection and positioning method. Figure 4 This is a structural block diagram of a wind turbine tower inspection and positioning device based on cluster partitioning, provided in an embodiment of this application. Figure 4 As shown, the inspection and positioning device for wind turbine towers based on cluster partitioning includes:

[0189] The model building module 41 is configured to build a three-dimensional geometric model of the wind turbine tower and to calibrate the physical reference points of the wind turbine tower. The physical reference points include at least a first reference point, which is set at the top and / or middle of the wind turbine tower.

[0190] The acquisition module 42 is configured to acquire the pose data of the UAV through the UAV's quantum inertial measurement unit;

[0191] The point cloud acquisition module 43 is configured to acquire point cloud data on the surface of the wind turbine tower using the lidar of a drone;

[0192] The point cloud matching module 44 is configured to perform point cloud matching on pose data based on point cloud data and three-dimensional geometric model to obtain point cloud matching results.

[0193] The pixel recognition module 45 is configured to acquire image data containing a first reference point through the gimbal camera of the drone, and determine the pixel coordinates of the first reference point in the image data.

[0194] The calibration module 46 is configured to perform reference point calibration on the pose data based on pixel coordinates to obtain the reference point calibration result;

[0195] The positioning module 47 is configured to locate the UAV based on pose data, point cloud matching results, and reference point calibration results, and obtain the UAV positioning result.

[0196] It should be noted that the model construction module 41, acquisition module 42, point cloud acquisition module 43, point cloud matching module 44, pixel recognition module 45, calibration module 46, and positioning module 47 mentioned above correspond to steps S201 to S207 in the embodiments. Multiple modules and their corresponding steps implement the same instances and application scenarios, but are not limited to the content disclosed in the above embodiments. It should also be noted that the above modules, as part of the device, can run on the computer terminal 10 provided in the embodiments.

[0197] Embodiments of this application may provide a computer device. Optionally, in this embodiment, the computer device may be located in at least one of a plurality of network devices in a computer network. The computer device includes a memory and a processor.

[0198] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the wind turbine tower inspection and positioning method and device in this embodiment. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby realizing the aforementioned wind turbine tower inspection and positioning method. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to a computer terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0199] The processor can call the information and application program stored in the memory through the transmission device to execute the above steps S201 to S207.

[0200] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a non-volatile storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0201] Embodiments of this application also provide a non-volatile storage medium. Optionally, in this embodiment, the aforementioned non-volatile storage medium can be used to store the program code executed by the wind turbine tower inspection and positioning method provided in the above embodiments.

[0202] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.

[0203] Embodiments of this application also provide a computer program product, including a computer program. Optionally, in this embodiment, when the computer program is executed by a processor, it can implement the methods of steps S201 to S207 described above.

[0204] In the above embodiments of this application, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0205] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0206] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0207] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0208] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a non-volatile storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0209] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of this application, and these improvements and modifications should also be considered within the scope of protection of this application.

Claims

1. A method for inspecting and locating the tower of a wind turbine, characterized in that, The method includes: A three-dimensional geometric model of the wind turbine tower is established, and the physical reference points of the wind turbine tower are marked, wherein the physical reference points include at least a first reference point, which is set at the top and / or middle of the wind turbine tower. The drone's pose data is obtained through the drone's quantum inertial measurement unit; The point cloud data of the wind turbine tower surface is collected by the lidar of the drone; Based on the point cloud data and the three-dimensional geometric model, point cloud matching is performed on the pose data to obtain point cloud matching results; The image data containing the first reference point is acquired by the gimbal camera of the drone, and the pixel coordinates of the first reference point in the image data are determined. Based on the pixel coordinates, the pose data is calibrated using reference points to obtain reference point calibration results. The UAV is located based on the pose data, the point cloud matching results, and the reference point calibration results to obtain the UAV positioning results.

2. The method according to claim 1, characterized in that, The establishment of the three-dimensional geometric model of the wind turbine tower includes: Based on the design parameters of the wind turbine tower and / or the initial flight scan data of the UAV, a three-dimensional geometric model is established. The wind turbine tower includes a tower body and blades. The geometric model parameters of the tower body include at least one of the cylindrical axis equation of the tower, the tower radius, and the tower height range. The geometric model parameters of the blades include the three-dimensional surface equation of the blades. The initial flight scan data includes the coordinate data of the physical reference point, which is obtained by measuring the coordinate data of the physical reference point through a positioning unit set on the UAV.

3. The method according to claim 1, characterized in that, The step of performing point cloud matching on the pose data based on the point cloud data and the three-dimensional geometric model to obtain point cloud matching results includes: The point cloud data is registered with the three-dimensional geometric model to obtain the point cloud registration result, and the first position information of the UAV is determined based on the point cloud registration result. The first position information is compared with the second position information in the pose data to determine the first positioning deviation of the UAV; If the first positioning deviation is greater than a preset deviation threshold, it is determined that the quantum inertial measurement unit has a positioning defect.

4. The method according to claim 1, characterized in that, The step of calibrating the pose data based on the pixel coordinates to obtain the reference point calibration result includes: The predicted coordinates of the first reference point are determined based on the pixel coordinates and the pose data of the UAV. The predicted coordinates of the first reference point are compared with the calibrated coordinates of the first reference point to determine the second positioning deviation of the quantum inertial measurement unit.

5. The method according to claim 1, characterized in that, The method further includes: Based on the image data, defect detection is performed on the wind turbine tower to identify tower defects and determine the pixel defect location of the tower defects. The pose data of the UAV, the ranging data of the lidar, and the gimbal attitude data of the gimbal camera corresponding to the tower defect are obtained, wherein the pose data includes the position data and attitude data of the UAV, and the ranging data is the distance from the UAV to the surface of the tower. The pixel defect location, pose data, ranging data, and gimbal attitude data are mapped onto the three-dimensional geometric model to determine the physical defect location of the tower defect.

6. The method according to claim 5, characterized in that, The step of mapping the pixel defect location, the pose data, the ranging data, and the gimbal attitude data onto the three-dimensional geometric model to determine the physical defect location of the tower defect includes: The position of the optical center of the gimbal camera is determined, wherein the position of the optical center is determined based on the pose data and the installation position between the optical center of the camera and the center of the UAV; The direction of the light ray pointing from the optical center of the camera toward the tower defect is determined based on the position of the camera's optical center. Construct ray parameter equations based on the ray direction; The equation of the tower's cylindrical surface is determined based on the aforementioned three-dimensional geometric model; Solving the ray parameter equation and the tower cylindrical surface equation yields the three-dimensional physical coordinates of the tower defect.

7. The method according to claim 6, characterized in that, Determining the direction of the light ray pointing from the camera's optical center to the tower defect based on the camera's optical center position includes: The pixel coordinates are converted into a normalized direction vector in the camera coordinate system, wherein the normalized direction vector is expressed as: ； In the formula, , The pixel coordinates of the drone. K This is the camera intrinsic parameter matrix; Transform the normalized direction vector to the world coordinate system to obtain the physical direction vector of the ray direction, wherein the physical direction vector is expressed as: ； In the formula, For the drone's attitude, Gimbal posture; The construction of ray parameter equations based on the ray direction includes: The ray parameter equation is constructed based on the camera optical center position and the physical direction vector, wherein the ray parameter equation is expressed as: ； In the formula, This is the position of the camera's optical center.

8. The method according to claim 6, characterized in that, After solving the ray parameter equation and the tower cylindrical surface equation to obtain the three-dimensional physical coordinates of the tower defect, the method further includes: The three-dimensional physical coordinates of the tower defect are converted into physical parameters in a cylindrical unfolded coordinate system, wherein the physical parameters include the defect height and the defect angle.

9. The method according to claim 6, characterized in that, After solving the ray parameter equation and the tower cylindrical surface equation to obtain the three-dimensional physical coordinates of the tower defect, the method further includes: Label the three-dimensional physical coordinates on the three-dimensional geometric model; and / or A tower inspection report is generated based on at least one of the following: physical defect location, defect size, defect type, and defect change trend.

10. A wind turbine tower inspection and positioning device based on cluster partitioning, characterized in that, include: The model building module is configured to build a three-dimensional geometric model of the wind turbine tower and to mark the physical reference points of the wind turbine tower, wherein the physical reference points include at least a first reference point, which is set at the top and / or middle of the wind turbine tower. The acquisition module is configured to acquire the pose data of the UAV through the UAV's quantum inertial measurement unit; The point cloud acquisition module is configured to acquire point cloud data of the surface of the wind turbine tower using the lidar of the UAV; The point cloud matching module is configured to perform point cloud matching on the pose data based on the point cloud data and the three-dimensional geometric model to obtain the point cloud matching result. The pixel recognition module is configured to acquire image data containing the first reference point through the gimbal camera of the drone, and determine the pixel coordinates of the first reference point in the image data. The calibration module is configured to perform reference point calibration on the pose data based on the pixel coordinates to obtain the reference point calibration result. The positioning module is configured to locate the UAV based on the pose data, the point cloud matching result, and the reference point calibration result, and obtain the UAV positioning result.

Images