Terahertz continuous wave wind turbine blade defect imaging method based on unmanned aerial vehicle platform

Abstract

The application discloses a terahertz continuous wave wind power blade defect imaging method based on a UAV platform, relates to the technical field of blade defect imaging, and solves the problems that, in the prior art, a crawling robot as a contact type mobile platform is used for detecting a wind power blade, the detection process is complicated, time-consuming and has safety risks, and the method provides an efficient and flexible non-contact detection mode: by using a UAV as an aerial autonomous bearing platform, the contact dependence on the surface of the blade is overcome, the rapid deployment of the detection equipment, long-distance maneuvering and flexible arrival are realized, and the detection operation efficiency and safety of the high-altitude and in-service wind power blade are greatly improved; and a split type terahertz transmission imaging method which can be integrated with the UAV platform is provided, the penetration characteristics of terahertz waves on composite materials are utilized, the defects such as bubbles and delamination in the blade are directly and high-resolution imaged, and the deficiency of the prior art in the detection capability of internal defects is made up.

Description

Technical Field

[0001] This invention relates to the field of blade defect imaging technology, specifically a terahertz continuous wave wind turbine blade defect imaging method based on an unmanned aerial vehicle (UAV) platform. Background Technology

[0002] As the wind power industry develops towards larger scale, the size of wind turbine blades continues to increase (exceeding 100 meters), and their health directly affects power generation efficiency and operational safety. The blades are mainly made of glass fiber reinforced composite materials, which are prone to internal defects during manufacturing and long-term operation. Currently, the mainstream industry method for high-altitude in-service inspection of wind turbine blades still relies on manual baskets or rope descents for close-range visual inspection and tapping. This method is highly dangerous, inefficient, and the inspection results are subjective and cannot quantify internal damage.

[0003] In the prior art, patent CN120156615A discloses an adaptive crawling robot for the surface of a wind turbine blade and its control system. The crawling robot consists of two parts: a mechanical structure and a control system. The mechanical structure includes a main structure and six underactuated leg structures. The control system consists of a finite state machine module, a trajectory calculation module, a control execution module, and a state perception module. It controls the robot by receiving instructions, sensing, adaptively adjusting, and moving.

[0004] However, as a contact-type mobile platform, the deployment of crawling robots requires the assistance of auxiliary hoisting equipment to install them onto the blade surface, a process that is cumbersome, time-consuming, and poses safety risks. The robot's movement speed on complex curved surfaces is slow, and its working range is limited to the starting position of a single deployment, making it difficult to quickly and efficiently complete a comprehensive survey of the entire blade or the entire wind farm. Furthermore, due to limitations in the robot's load capacity, size, and stability on vertical walls, the system is difficult to integrate relatively precise heavy-duty non-destructive testing equipment, such as terahertz imagers, which have requirements for installation posture.

[0005] To address the aforementioned technical shortcomings, a solution is proposed. Summary of the Invention

[0006] The purpose of this invention is to solve the problems mentioned above by proposing a terahertz continuous wave wind turbine blade defect imaging method based on an unmanned aerial vehicle (UAV) platform; and to provide an efficient and flexible non-contact inspection method: by using UAVs as an autonomous aerial platform, the dependence on contact with the blade surface is overcome, enabling rapid deployment, long-distance mobility and flexible arrival of inspection equipment, which greatly improves the efficiency and safety of inspection operations on high-altitude and in-service wind turbine blades.

[0007] Furthermore, a split-type terahertz transmission imaging method that can be integrated with an unmanned aerial vehicle platform is provided. By utilizing the penetrating characteristics of terahertz waves through composite materials, intuitive and high-resolution imaging of defects such as bubbles and delamination inside blades can be achieved, thus making up for the shortcomings of existing technologies in detecting internal defects.

[0008] The objective of this invention can be achieved through the following technical solutions:

[0009] A terahertz continuous wave wind turbine blade defect imaging method based on an unmanned aerial vehicle (UAV) platform is as follows:

[0010] Step 1: System initialization and task preparation; power on the device and perform a communication self-test.

[0011] Step 2: Automatic flight and detection point ready, plan 3D flight detection route; the UAV flies to the preset detection point in coordination, and automatically adjusts to the relative position in the air to make stable hovering decisions;

[0012] Step 3: Trigger command is generated. After confirming that the hovering is stable, transient interference compensation analysis is performed, and a trigger command is generated and sent to the host computer software.

[0013] Step 4: Real-time coordinate acquisition and automatic data container creation; After receiving the trigger command, the host computer software sends a one-time coordinate request command to the UAV flight control system; The flight control system returns a data packet in real time; An independent folder directly named with the corresponding coordinate string is automatically created;

[0014] Step 5: Synchronous acquisition, processing, and storage of terahertz data; While requesting coordinates, the terahertz receiving unit continuously acquires terahertz wave signals penetrating the blade, generating a raw data stream; this data is repackaged and transmitted to the ground detector's host computer software; upon receiving the data stream, the host computer software performs real-time imaging processing:

[0015] Step 6: Single-point acquisition ends and loops; after acquisition is complete, stop writing data to the current coordinate folder and notify the drone to fly to the next preset detection point; the system repeats steps 2 to 6 until all points have been acquired;

[0016] Step 7: Task completed and structured data ready; the folder named with coordinate strings in the task root directory stores all image sequences corresponding to the spatial locations;

[0017] Step 8: Intelligent Analysis and Report Generation; Perform image analysis on the grayscale image sequence and generate a report.

[0018] Furthermore, the design of the terahertz transmitting and receiving units within the airborne detection subsystem is as follows: the unit shell is made of magnesium-aluminum alloy CNC machined, and the surfaces of the internal structural components are all treated with black anodized matte finish to form an extremely black inner cavity to absorb stray reflected light.

[0019] Furthermore, the process of making a stable hovering decision is as follows:

[0020] The UAVs containing the terahertz transmitting unit and the terahertz receiving unit are respectively marked as the transmitting flight end and the receiving flight end; according to the detection time of the preset detection point, data collection and analysis are performed on the environment of the flight area at the detection time.

[0021] The data acquisition area is determined based on the area where the preset detection point is located, and atmospheric parameters within the data acquisition area are obtained, specifically atmospheric visibility, ambient wind speed, and real-time wind direction. Based on the current atmospheric parameters, the communication parameters of the transmitting and receiving terminals corresponding to the current detection time are determined, specifically the communication delay duration and communication delay frequency.

[0022] The detection time is set as the center of the detection time window. Based on the atmospheric parameters and communication parameters in the data collection area corresponding to the detection time window, the features covered by the parameters are synchronously identified and detected. If the parameter features change, they are marked as risk feature vectors; otherwise, if the parameter features do not change, they are marked as safe feature vectors.

[0023] Furthermore, the risk feature vectors are weighted and fused to output the environmental level within the current data acquisition area. Based on the comparison of environmental levels, a decision is made regarding the terahertz emission to be performed within the current pre-detection time window. If the environmental level is lower than the set threshold, the current preset detection point is detected, and within the current pre-detection time window, a stable hovering attempt is made based on the interval between the current time and the detection time, with the current detection time adjusted according to the progress of the stable hovering. Conversely, if the environmental level is not lower than the set threshold, no action is taken, i.e., the preset detection point is replaced or the detection time is adjusted.

[0024] Furthermore, when the current environmental level is not lower than the set level threshold, a post-detection time window is set based on the detection time, and the duration of the post-detection time window is consistent with that of the pre-detection time window. The environmental level is calculated based on the data fluctuation prediction of atmospheric parameters and communication parameters in the data acquisition area, and the environmental level is obtained according to the predicted parameter coverage characteristics. If the deviation between the preset environmental level and the set level threshold exceeds the deviation value, the predicted detection point is changed; otherwise, if the deviation between the preset environmental level and the set level threshold does not exceed the deviation value, the detection time of the current preset detection point is adjusted.

[0025] Furthermore, the process of transient disturbance compensation analysis is as follows:

[0026] The wind speed change span of the corresponding areas of the transmitting and receiving ends is obtained, and the wind speed change span difference is calculated based on the difference and marked as the inter-end wind speed difference; and the airflow change spacing on the corresponding end surfaces is obtained based on the transmitting and receiving ends, and the airflow change spacing difference is calculated based on the difference and marked as the inter-end airflow spacing difference.

[0027] If the wind speed difference between the terminals exceeds the wind speed difference threshold, or the airflow distance difference between the terminals exceeds the distance difference threshold, it is inferred that there is a risk of imbalance between the transmitting and receiving terminals. The current moment is marked as a transient interference moment, and transient interference compensation is performed. If the wind speed difference between the terminals does not exceed the wind speed difference threshold, and the airflow distance difference between the terminals does not exceed the distance difference threshold, it is inferred that there is no risk of imbalance between the transmitting and receiving terminals. The current moment is marked as a transient interference-free moment.

[0028] Furthermore, the image analysis process is as follows:

[0029] The grayscale image sequence is determined, and the images are extracted for image recognition analysis. Based on the historical maintenance process, the image brightness corresponding to the wind turbine blade defects is determined, and the image brightness threshold corresponding to the defects is determined.

[0030] The extracted image is divided into equal regions, and the brightness of each region is extracted and compared with the image brightness threshold.

[0031] If the brightness of a region does not exceed the image brightness threshold, the corresponding region will be marked as a low-brightness region.

[0032] If the brightness of a region exceeds the image brightness threshold, the corresponding region will be marked as a highlighted region.

[0033] Furthermore, defect locations are marked according to region type;

[0034] Simultaneously, the brightness type of each region in the image is extracted and labeled, and a brightness distribution map is constructed. Based on the grayscale image sequence, a brightness distribution map set is obtained, the regions in the brightness distribution map set where the type changes are obtained, and the defect development trajectory is constructed based on the region connection.

[0035] If the number of regions within the defect development trajectory continues to increase, the current defect development trajectory will be marked as a dynamic defect trajectory and uploaded synchronously, and maintenance control will be performed synchronously based on the dynamic defect trajectory; if the number of regions within the defect development trajectory does not continue to increase, the current defect development trajectory will be marked as a static defect trajectory and uploaded synchronously, and maintenance will be performed based on the location of the dynamic defect trajectory.

[0036] Compared with the prior art, the beneficial effects of the present invention are:

[0037] 1. Significantly improved detection efficiency and safety: This invention uses drones as a non-contact aerial platform, eliminating the complex hoisting and adsorption process required by crawling robots, and realizing rapid deployment and flexible mobility of detection equipment; drones can autonomously fly to any position on the blade, which is especially suitable for rapid general inspection or targeted review of high-altitude, in-service blades, and the work efficiency and personnel safety are far superior to contact solutions.

[0038] 2. Achieves high-precision direct imaging of internal defects in blades: This invention adopts split-type terahertz transmission imaging technology, which utilizes the penetrability of terahertz waves to composite materials to directly obtain energy attenuation images reflecting the uniformity of internal materials. Compared with the ultrasonic (requires coupling agent) or visual (surface only) inspection mainly relied upon by crawling robots, this invention can intuitively and effectively discover key internal defects such as bubbles and delamination, and the detection information is more comprehensive and direct.

[0039] 3. Seamless, accurate, and automated association between detection data and absolute spatial location: Unlike existing technologies that rely on indirect and error-prone methods of "encoder calculation," this invention automatically acquires the high-precision RTK geographic coordinates of the UAV at the moment of acquisition and directly names and creates data folders based on these coordinates. This solution achieves a native one-to-one binding between data and real-world coordinates, fundamentally eliminating the problems of cumulative errors and model dependence, so that each frame of data has traceable absolute position information, greatly improving the reliability of detection results and the convenience of subsequent analysis.

[0040] 4. During the hovering phase, environmental data collection and analysis improve the accuracy of hovering environment identification, thereby reducing the number of hovering attempts and avoiding unnecessary flight risks. At the same time, when hovering is not decided, data analysis is used to further decide whether to change the preset detection point or adjust the detection time, which improves the overall efficiency of the detection process.

[0041] 5. When the trigger command is generated, both the transmitting and receiving flight ends hover continuously. To ensure normal transmission and reception of terahertz waves, transient interference compensation is required during the hovering phase. Based on wind speed and direction data, the motor speed is adjusted in real time through flight control algorithms to counteract wind load torque. The thrust of the motor on the upwind side is increased to prevent the UAV from being blown away from the hovering point. When the two aircraft cooperate, wake interference needs to be considered, and the impact of wake interference is reduced by adjusting the relative position or power compensation.

[0042] By using barometer and lidar data, the error caused by air pressure changes with temperature is eliminated, and the accurate height relative to the ground is output. For slopes and uneven ground, the ground plane features are extracted by a vision camera to correct the hovering height reference. Attached Figure Description

[0043] To facilitate understanding by those skilled in the art, the present invention will be further described below with reference to the accompanying drawings.

[0044] Figure 1 This is a flowchart of the method of the present invention;

[0045] Figure 2 This is a schematic diagram of an actual implementation scenario of the present invention. Detailed Implementation

[0046] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0048] This application proposes an aerial automated inspection system that can reach the detection position non-contactly, carry precision physical sensing equipment, and realize automatic and high-precision binding of detection data with the absolute spatial position of the blade. The UAV platform serves as the system execution hub, and the entire system includes an aerial detection subsystem, a ground control subsystem, and a network communication subsystem.

[0049] The hardware of the aerial detection subsystem consists of two industrial-grade UAV platforms, each carrying a terahertz transmitting unit and a terahertz receiving unit, forming an aerial carrier for split-type transmission imaging.

[0050] The hardware of the ground control subsystem consists of a high-performance ground workstation computer, on which the detector host computer software and report generation software run;

[0051] The network communication subsystem includes a portable high-throughput WiFi 6 base station, which establishes a stable TCP / IP data channel between the air unit and the ground unit.

[0052] Example 1

[0053] The aerial detection subsystem, ground control subsystem, and network communication subsystem work together to image wind turbine blade defects using terahertz continuous wave technology. (See also: [link to method description]) Figure 1 - Figure 2As shown, a terahertz continuous wave wind turbine blade defect imaging method based on an unmanned aerial vehicle (UAV) platform is described below:

[0054] Step 1: System initialization and task preparation;

[0055] The operator powers on all equipment and starts the detector host computer software on the ground workstation; the software automatically establishes a stable TCP / IP connection with the STM32 main control board of the air unit and the UAV flight control system through a portable WiFi base station and completes the communication self-test;

[0056] Step 2: Automatic flight and detection point ready;

[0057] The operator executes a pre-planned three-dimensional flight inspection route for the target blade in the UAV professional ground station software; two UAVs take off automatically, fly together to the first preset inspection point, and automatically adjust to their relative positions in the air to achieve stable hovering, so that the terahertz beam can penetrate the area to be tested on the blade vertically;

[0058] Step 3: Manual confirmation and data collection triggering;

[0059] After the operator observes the status parameters in the software interface and confirms that the status is stable, they manually click the "Start Recording" button on the detector's host computer software interface to generate a trigger command and send it to the host computer software.

[0060] Step 4: Real-time coordinate acquisition and automatic creation of data containers;

[0061] As the key to achieving automatic binding of data and spatial location, after receiving the trigger command, the host computer software immediately sends a one-time coordinate request command to the UAV flight control system through the established TCP link; the flight control system returns a data packet in real time in the format of "N, X, Y, Z", where N is the point number, X and Y are the high-precision latitude and longitude coordinates on the horizontal plane (from the RTK-GNSS module), and Z is the altitude.

[0062] Subsequently, the software immediately creates a unique folder in the root directory of this detection task preset in the local file system, named directly with the coordinate string, for example, 3_121.1234_31.5678_150.5; this folder is unique for all data of this spatial point.

[0063] Step 5: Synchronous acquisition, processing and storage of terahertz data;

[0064] While requesting coordinates, the 64×64 pixel array detector in the terahertz receiving unit begins to continuously collect terahertz wave signals after penetrating the blade, generating a raw data stream. This data stream is transmitted to the STM32 main control board via the USBCDC protocol. The main control board does not perform complex processing; its main function is to act as a "protocol conversion bridge," repackaging the USB data stream into TCP / IP data packets and stably transmitting them back to the ground detector's host computer software via the WiFi 6 module.

[0065] After receiving the data stream, the host computer software performs real-time imaging processing: first, it reconstructs the one-dimensional data stream into a 64×64 two-dimensional pixel matrix; then, it performs rapid temporal filtering, flat field correction, and other processing to generate a terahertz transmission grayscale image sequence for that point (e.g., 30 frames per second); these image frames are saved one by one in real time to the coordinate folder corresponding to this point created in step 4, named image_001.png, image_002.png, etc.; the software interface synchronously displays the dynamic terahertz video stream for the operator to monitor the detection quality in real time.

[0066] Step 6: End of single-point data collection and loop;

[0067] Once the operator determines that sufficient data has been collected at a given location based on the real-time video, they click the "Stop Recording" button. The software stops writing data to the current coordinate folder and instructs the drone to fly to the next preset detection point. The system repeats steps 2 to 6 until all locations have been collected.

[0068] Step 7: Task completed and structured data ready;

[0069] After all locations were collected, the drone automatically returned and landed. At this point, the data was perfectly structured in the local file system: the task root directory contained a series of folders named "N_X_Y_Z", and each folder stored the complete image sequence for the corresponding spatial location. Throughout the process, the association between the data and the absolute geographic coordinates was automatically completed at the moment of collection, without any post-processing or manual intervention.

[0070] Step 8: Intelligent analysis and report generation;

[0071] After the drone lands, the operator opens a browser to access the locally running report generation software (a web application). Within the software interface, the operator selects the root directory for the current inspection task. The software automatically scans and lists all coordinate folders; the operator can flexibly select (e.g., filter by entire blades or customize selections) the inspection points to be analyzed; the software directly reads images from the corresponding coordinate folders, providing analysis tools; finally, with a single click, the operator can generate a standardized PDF inspection report integrating all selected point information, defect images, and analysis conclusions, which can be automatically uploaded to the cloud for archiving.

[0072] Example 2

[0073] In the previous embodiment, the design of the terahertz transmitting unit and terahertz receiving unit in the airborne detection subsystem was as follows: the unit shell was made of magnesium-aluminum alloy CNC machined to achieve IP64 protection level; the internal core (terahertz source, detector, lens group) was rigidly fixed by a precision bracket.

[0074] Crucially, all internal structural components undergo a black anodized matte finish to create an "extremely black" interior, maximizing the absorption of stray reflected light and reducing background noise.

[0075] Shock absorption and attitude stabilization design: The unit is connected to the connecting plate via a top shock-absorbing base (with built-in silicone shock-absorbing pads) and then mounted to the drone; ensuring stable beam direction during flight vibrations.

[0076] The core imaging process of the specific airborne detection subsystem is as follows: a 110GHz solid-state source generates a continuous wave → the lens collimates it into a parallel beam → it penetrates the blade vertically → the receiving end lens focuses the beam → the 64×64 array detector converts the optical signal into an electrical signal → after onboard amplification and ADC conversion, the raw digital signal stream is output through the USB interface.

[0077] Specifically: The transmitting optical path: The continuous wave generated by the 110GHz solid-state terahertz source first passes through a collimating lens made of high-resistivity silicon or high-density polyethylene. This lens is custom-designed to shape the diverging beam into a parallel beam with a diameter matching the preset detection area. This parallel beam passes through the high-density polyethylene optical window at the front end of the transmitting unit and enters the wind turbine blade inspection area at a vertical or near-vertical angle.

[0078] Penetration and Interaction: Terahertz parallel beams penetrate the composite material of the blade. In normal regions, the beam undergoes weak absorption and scattering with minimal energy attenuation; in defective regions (such as bubbles or delamination), due to abrupt changes in the dielectric constant, the beam experiences strong scattering, reflection, and energy absorption.

[0079] Receiving optical path: The outgoing beam carrying the internal information of the blade (i.e., energy attenuation distribution) enters the receiving unit; first, it passes through the optical window of the receiving end, and then a set of focusing lenses (material as before) focuses the penetrating terahertz beam onto the sensitive surface of the 64×64 pixel array terahertz detector.

[0080] Example 3

[0081] When the first embodiment is executed, the airborne detection subsystem is further improved during stable hovering. During the hovering phase, environmental data is collected and analyzed to improve the accuracy of hovering environment identification, thereby reducing the number of hovering attempts and avoiding unnecessary flight risks.

[0082] The UAVs containing the terahertz transmitting unit and the terahertz receiving unit are respectively marked as the transmitting flight end and the receiving flight end; based on the detection time pairs of the preset detection points, data collection and analysis are performed on the environment of the flight area during the detection time.

[0083] The data acquisition area is determined based on the area where the preset detection point is located, and atmospheric parameters within the data acquisition area are obtained, specifically atmospheric visibility, ambient wind speed, and real-time wind direction. Based on the current atmospheric parameters, the communication parameters of the transmitting and receiving terminals corresponding to the current detection time are determined, specifically the communication delay duration and communication delay frequency.

[0084] The detection time is set as the center of the detection time window. Based on the atmospheric parameters and communication parameters in the data collection area corresponding to the detection time window, the features covered by the parameters are identified and detected synchronously. If the parameter features change, they are marked as risk feature vectors; otherwise, if the parameter features do not change, they are marked as safe feature vectors.

[0085] The risk feature vectors are weighted and fused to output the environmental level within the current data collection area. Based on the comparison of environmental levels, a decision is made regarding the terahertz emission to be performed within the current pre-detection time window. If the environmental level is lower than the set threshold, the current preset detection point is detected, and within the current pre-detection time window, a stable hovering attempt is made based on the interval between the current time and the detection time, with the current detection time adjusted according to the progress of the stable hovering. Conversely, if the environmental level is not lower than the set threshold, no action is taken, i.e., the preset detection point is replaced or the detection time is adjusted.

[0086] As a further solution of the present invention, when the current environmental level is not lower than the set level threshold, a post-detection time window is set according to the detection time, and the post-detection time window is the same as the pre-detection time window in duration; the environmental level is calculated based on the data fluctuation prediction of atmospheric parameters and communication parameters in the data acquisition area, and the environmental level is obtained according to the predicted parameter coverage characteristics. If the deviation between the preset environmental level and the set level threshold exceeds the deviation value, the predicted detection point is changed; otherwise, if the deviation between the preset environmental level and the set level threshold does not exceed the deviation value, the detection time of the current preset detection point is adjusted.

[0087] Example 4

[0088] Based on the execution of Embodiment 1 and Embodiment 3, when the trigger command is generated, the transmitting and receiving flight ends continue to hover. In order to ensure the normal transmission and reception of terahertz waves, transient interference compensation is required during the hovering phase.

[0089] Therefore, this embodiment performs transient interference compensation analysis based on the previous embodiment;

[0090] The wind speed change span of the corresponding areas of the transmitting and receiving ends is obtained, and the wind speed change span difference is calculated based on the difference and marked as the inter-end wind speed difference; and the airflow change spacing on the corresponding end surfaces is obtained based on the transmitting and receiving ends, and the airflow change spacing difference is calculated based on the difference and marked as the inter-end airflow spacing difference.

[0091] If the wind speed difference between the terminals exceeds the wind speed difference threshold, or the airflow distance difference between the terminals exceeds the distance difference threshold, it is inferred that there is a risk of imbalance between the transmitting and receiving terminals. The current moment is marked as the moment of transient interference, and transient interference compensation is performed.

[0092] If the wind speed difference between the terminals does not exceed the wind speed difference threshold and the airflow distance difference between the terminals does not exceed the distance difference threshold, it is inferred that there is no risk of imbalance between the transmitting and receiving terminals, and the current moment is marked as a transient interference-free moment.

[0093] It is understandable that transient disturbance compensation is expressed as:

[0094] Based on wind speed and direction data, the motor speed is adjusted in real time through flight control algorithms to counteract wind load torque; for example, the thrust of the motor on the upwind side is increased to prevent the drone from being blown away from the hovering point; when two drones work together, wake interference needs to be considered (the downwash airflow of the front drone will affect the stability of the rear drone), and the wake effect can be reduced by adjusting the relative position (such as the height difference between the front and rear drones ≥ 1.5 times the rotor diameter) or by power compensation.

[0095] By using barometer and lidar data, the error caused by air pressure changes with temperature is eliminated, and the accurate altitude relative to the ground (rather than absolute altitude) is output; for slopes and uneven ground, the ground planar features are extracted by a visual camera to correct the hovering height reference.

[0096] Example 5

[0097] According to the above embodiments, after completing the terahertz wave transmission and reception process at multiple preset detection points, image analysis is performed based on the grayscale image sequence corresponding to each geographical coordinate to facilitate defect imaging analysis and generate a defect report; by utilizing the penetrating characteristics of terahertz waves on composite materials, intuitive and high-resolution imaging of defects such as bubbles and delamination inside the blades can be achieved, making up for the shortcomings of existing technologies in detecting internal defects.

[0098] The grayscale image sequence is determined, and the images are extracted for image recognition analysis. Based on the historical maintenance process, the image brightness corresponding to the wind turbine blade defects is determined, and the image brightness threshold corresponding to the defects is determined.

[0099] The extracted image is divided into equal regions, and the brightness of each region is extracted and compared with the image brightness threshold.

[0100] If the brightness of a region does not exceed the image brightness threshold, the corresponding region will be marked as a low-brightness region.

[0101] If the brightness of a region exceeds the image brightness threshold, the corresponding region will be marked as a highlighted region.

[0102] It needs to be explained that normal areas are characterized by uniform material and stable attenuation, resulting in continuous high-brightness areas (bright areas) in the image; defective areas, such as bubbles and delamination, cause abrupt changes in the dielectric constant, leading to strong scattering or reflection, which drastically reduces the penetration energy, resulting in isolated low-brightness areas (dark areas) in the image.

[0103] Mark defect locations according to region type;

[0104] Simultaneously, the brightness type of each region in the image is extracted and labeled, and a brightness distribution map is constructed. Based on the grayscale image sequence, a brightness distribution map set is obtained, the regions in the brightness distribution map set where the type changes are obtained, and the defect development trajectory is constructed based on the region connection.

[0105] If the number of regions within the defect development trajectory continues to increase, the current defect development trajectory will be marked as a dynamic defect trajectory and uploaded synchronously. Maintenance and control will be performed synchronously based on the dynamic defect trajectory.

[0106] If the number of regions within the defect development trajectory does not continue to increase, the current defect development trajectory will be marked as a static defect trajectory and uploaded synchronously, and maintained according to the location of the dynamic defect trajectory.

[0107] Example 6

[0108] This implementation makes technical substitutions based on the above embodiments, with the drone platform being a substitute: the core "aerial autonomous mobile platform" is not limited to multi-rotor drones; tethered drones (which provide continuous power and data transmission via cables and can achieve ultra-long flight time), unmanned helicopters (which have higher load capacity and wind resistance) or compound-wing drones (which combine vertical take-off and landing with fixed-wing cruise and are suitable for large-scale wind field inspections) can all be used as substitute vehicles to carry terahertz detection units;

[0109] Platform quantity and configuration alternatives: The split "transmit-receive" dual-machine mode is the preferred solution; however, to achieve the same invention purpose, a single-platform configuration can also be adopted, for example: on a large UAV, the transmitting unit and the receiving unit are placed on both sides of the blade through a retractable robotic arm or pod; or a reflective single-end detection is adopted, that is, only a UAV equipped with a terahertz transceiver module is used to image the blade by receiving the echo reflected / scattered by internal defects; although the detection mode is different, this solution is also based on the UAV platform to achieve non-contact detection of high-altitude blades.

[0110] Example 7

[0111] Alternative solutions are provided for data storage, including alternatives to the data container creation logic: the core of associating detection data with spatial coordinates is not limited to "creating a physical folder"; equivalent solutions include: when starting acquisition, directly writing the acquired coordinate information as metadata tags into the header of each frame image file (such as EXIF ​​information) or a separate index file; or creating a record in the database with coordinates as the primary key index and associating all image data of that point as a binary large object or file path with it; alternatives to the timing of triggering association: in addition to requesting coordinates and creating associations when "starting recording", pre-association or continuous association can also be used;

[0112] For example, during the flight of a drone, high-frequency coordinate trajectories and timestamps are continuously recorded, while a terahertz detector continuously collects data and timestamps it. Afterwards, the data stream and coordinate trajectories are matched and aligned through time synchronization to achieve correlation.

[0113] Example 8

[0114] Provide alternatives to communication and data transmission methods.

[0115] The data link between the air unit and the ground unit is not limited to WiFi 6; using 5G private networks (deployed in wind farms), private microwave relays, or lightweight tethered optical cables (in tethered drone solutions) to achieve high-speed and stable data transmission are all feasible alternatives.

[0116] Airborne main control and protocol conversion replacement: The "protocol conversion bridge" function undertaken by the STM32 main control board can be implemented by embedded processors of other architectures (such as Raspberry Pi computing modules, RK series chips, etc.); in terms of data protocol, the detector output can also be industrial camera protocols such as GigEVision and CameraLink, which can be connected to the communication link through the corresponding converter.

[0117] In summary, this invention discloses a defect imaging method for wind turbine blades, in which:

[0118] The automatic association and storage method based on real-time geographic coordinates of UAVs, with its automated "request-create-store" process, achieves an inherent connection between detection data and spatial location, which is the core solution to data management pain points. The structure of the split-type UAV-borne terahertz transmission imaging system forms the physical basis for non-contact, bilateral, transmission-type internal imaging of large, in-service wind turbine blades. The broadband light-absorbing structure inside the terahertz detection unit: to suppress stray light interference, a black anodizing matte finish is applied to the inner wall of the terahertz unit's outer shell and all structural components, including the internal support, and its application in this specific scenario.

[0119] Thresholds, preset values, preset ranges, etc. are set for result comparison and analysis to determine whether they are good or bad. The value of these thresholds is determined by a combination of large-scale model analysis of sample data and human experience. They can also be adjusted appropriately based on seasonal or common-sense influences.

[0120] Furthermore, the settings for weighting ratios, influence factors, etc., are based on the magnitude of each parameter's influence on the results. The specific values ​​are allocated to ultimately reflect the impact on the results. The settings for input and storage are also determined by a combination of large-scale model analysis of sample data and human experience. Appropriate adjustments can also be made based on seasonal or rational influence conditions.

[0121] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A terahertz continuous wave wind turbine blade defect imaging method based on an unmanned aerial vehicle (UAV) platform, characterized in that, The defect imaging method is as follows: Step 1: Perform a communication self-test on the drone platform; Step 2: After communication is successful, the preset detection point is determined according to the real-time detection task, and the UAV flight path is determined based on the preset detection point. When the UAV arrives at the preset detection point, a stable hovering decision is made. During the hovering phase, parameter feature marking is performed through environmental data collection and analysis. The environmental level is obtained by weighted fusion based on the parameter feature marking type. The hovering decision is made through environmental level analysis. Step 3: After the hovering decision is determined, the timing of transient interference is determined based on real-time wind speed data collection and analysis, and transient interference compensation is performed according to the timing type; the transient interference compensation analysis process is as follows: Obtain the span of wind speed change in the corresponding areas of the transmitting and receiving terminals, calculate the difference in wind speed change span based on the difference, and mark it as the inter-terminal wind speed difference; Based on the transmitting and receiving ends, the distance between the airflow changes on the corresponding end surfaces is obtained. The difference in the airflow change distance is calculated and marked as the airflow distance difference between the ends. If the wind speed difference between the terminals exceeds the wind speed difference threshold, or the airflow distance difference between the terminals exceeds the distance difference threshold, it is inferred that there is a risk of imbalance between the transmitting and receiving terminals. The current moment is marked as the moment of transient interference, and transient interference compensation is performed. If the wind speed difference between the terminals does not exceed the wind speed difference threshold and the airflow distance difference between the terminals does not exceed the distance difference threshold, it is inferred that there is no risk of imbalance between the transmitting and receiving terminals, and the current moment is marked as a transient interference-free moment. Step 4: After the hovering phase decision is completed, determine the position coordinates and perform real-time imaging processing: Step 5: After completing the imaging process, establish the image sequence, perform image analysis simultaneously, and generate a report.

2. The terahertz continuous wave wind turbine blade defect imaging method based on an unmanned aerial vehicle platform according to claim 1, characterized in that, The design of the terahertz transmitting and receiving units in the airborne detection subsystem is as follows: the unit shell is made of magnesium-aluminum alloy CNC machined, and the surface of the internal structural components is treated with black anodized matte finish to form an extremely black inner cavity to absorb stray reflected light.

3. The terahertz continuous wave wind turbine blade defect imaging method based on an unmanned aerial vehicle platform according to claim 2, characterized in that, The process of making a stable hovering decision is as follows: The UAVs containing the terahertz transmitting unit and the terahertz receiving unit are respectively marked as the transmitting flight end and the receiving flight end; according to the detection time of the preset detection point, data collection and analysis are performed on the environment of the flight area at the detection time. The data acquisition area is determined based on the area where the preset detection point is located, and atmospheric parameters within the data acquisition area are obtained, specifically atmospheric visibility, ambient wind speed, and real-time wind direction. Based on the current atmospheric parameters, the communication parameters of the transmitting and receiving terminals corresponding to the current detection time are determined, specifically the communication delay duration and communication delay frequency. The detection time is set as the center of the detection time window. Based on the atmospheric parameters and communication parameters in the data collection area corresponding to the detection time window, the features covered by the parameters are synchronously identified and detected. If the parameter features change, they are marked as risk feature vectors; otherwise, if the parameter features do not change, they are marked as safe feature vectors.

4. The terahertz continuous wave wind turbine blade defect imaging method based on an unmanned aerial vehicle platform according to claim 3, characterized in that, The risk feature vectors are weighted and fused to output the environmental level within the current data collection area. Based on the comparison of environmental levels, a decision is made on whether to perform terahertz emission in the current pre-detection time window. If the environmental level is lower than the set threshold, the current preset detection point is detected, and within the current pre-detection time window, a stable hovering attempt is made based on the interval between the current time and the detection time, and the current detection time is adjusted according to the progress of the stable hovering. Conversely, if the environmental level is not lower than the set threshold, no action is taken, i.e., the preset detection point is replaced or the detection time is adjusted.

5. The terahertz continuous wave wind turbine blade defect imaging method based on an unmanned aerial vehicle platform according to claim 4, characterized in that, When the current environmental level is not lower than the set level threshold, a post-detection time window is set according to the detection time, and the post-detection time window is the same as the pre-detection time window in duration; the environmental level is calculated based on the data fluctuation prediction of atmospheric parameters and communication parameters in the data collection area, and the environmental level is calculated based on the predicted parameter coverage characteristics to obtain the preset environmental level; if the preset environmental level deviates from the set level threshold by more than the deviation value, the predicted detection point is replaced. Conversely, if the deviation between the preset environmental level and the set level threshold does not exceed the deviation value, the detection time of the current preset detection point will be adjusted.

6. The terahertz continuous wave wind turbine blade defect imaging method based on an unmanned aerial vehicle platform according to claim 1, characterized in that, The image analysis process is as follows: The grayscale image sequence is determined, and the images are extracted for image recognition analysis. Based on the historical maintenance process, the image brightness corresponding to the wind turbine blade defects is determined, and the image brightness threshold corresponding to the defects is determined. The extracted image is divided into equal regions, and the brightness of each region is extracted and compared with the image brightness threshold. If the brightness of a region does not exceed the image brightness threshold, the corresponding region will be marked as a low-brightness region. If the brightness of a region exceeds the image brightness threshold, the corresponding region will be marked as a highlighted region.

7. The terahertz continuous wave wind turbine blade defect imaging method based on an unmanned aerial vehicle platform according to claim 6, characterized in that, Mark defect locations according to region type; Simultaneously, the brightness type of each region in the image is extracted and labeled, and a brightness distribution map is constructed. Based on the grayscale image sequence, a brightness distribution map set is obtained, the regions in the brightness distribution map set where the type changes are obtained, and the defect development trajectory is constructed based on the region connection. If the number of regions within the defect development trajectory continues to increase, the current defect development trajectory will be marked as a dynamic defect trajectory and uploaded synchronously, and maintenance control will be performed synchronously based on the dynamic defect trajectory; if the number of regions within the defect development trajectory does not continue to increase, the current defect development trajectory will be marked as a static defect trajectory and uploaded synchronously, and maintenance will be performed based on the location of the dynamic defect trajectory.

Images