A smart inspection robot for wind turbine blades and its control method

Abstract

This invention discloses an intelligent inspection robot for wind turbine blades and its control method, belonging to the field of wind power equipment operation and maintenance technology. The robot includes a main hardware body, a multi-view inspection module, a gimbal assembly, a central control unit, and a power supply module. The main hardware body meets IP54 protection level and features a lightweight design, adaptable to operation in the narrow space of the blade cavity. The multi-view inspection module integrates an ultrasonic adaptive dust removal unit to achieve clear imaging in all dimensions. The control method constructs a complete closed-loop system of "digital twin base - adaptive closed-loop inspection - multimodal defect identification - twin fusion data processing - hierarchical emergency handling," supporting intelligent switching of inspection modes, accurate defect identification, and full life-cycle operation and maintenance analysis. This invention solves the problems of poor imaging quality, high defect omission rate, inability to balance inspection efficiency and accuracy, and insufficient safety protection reliability in traditional blade inspection, significantly improving the intelligence level and operational safety of wind turbine blade operation and maintenance.

Description

Technical Field

[0001] This invention relates to the field of wind power equipment operation and maintenance technology, and in particular to an intelligent inspection robot for wind turbine blades and its control method. Background Technology

[0002] As the core component for energy capture, wind turbine blades are exposed to harsh outdoor environments for extended periods. The blade cavity is prone to defects such as bulging, wrinkling, delamination, and cracks due to alternating loads, wind and sand erosion, and temperature changes. If these defects are not detected and addressed in a timely manner, they may lead to serious safety accidents such as blade breakage.

[0003] The current technical challenges in wind turbine blade cavity inspection are mainly as follows: (1) Poor hardware compatibility: The existing inspection equipment is too large and heavy, making it difficult to adapt to the narrow and curved space of the blade cavity. In the dusty environment of the blade cavity, the lens is prone to contaminants, resulting in blurred images and missed defects. Fixed-level supplementary lighting is prone to overexposure / underexposure problems, which cannot meet the imaging needs of complex environments.

[0004] (2) Insufficient accuracy of defect identification: Existing technologies mostly use general visual recognition algorithms, without adapting to the specific characteristics of the blade structure. They rely solely on single visible light image recognition, which cannot detect hidden defects under the coating. The accuracy of identification of the four types of core defects is generally below 95%, with prominent issues of misjudgment and missed judgment.

[0005] (3) Rigid inspection control logic: The existing inspection mode relies on manual switching between semi-automatic and manual modes. Fixed parameter inspection cannot balance efficiency and accuracy. The inspection efficiency of defect-free areas is low, and the detection of suspected defect areas is insufficient, making it impossible to achieve closed-loop control that adapts to working conditions.

[0006] (4) Low reliability of safety protection: The existing safety protection is a passive action triggered by a single threshold. After the communication is interrupted, the inspection data is easily lost and the task is interrupted. There is no adaptive adjustment during the rope retraction process, which can easily cause the robot to collide with the inner wall of the blade cavity. The anti-collision logic that stops the robot when there is slight contact seriously affects the inspection efficiency.

[0007] (5) Insufficient depth of data application: The existing data processing only realizes basic image stitching and defect classification statistics, without deep integration with blade structure, making it impossible to accurately trace the source of defects and predict development trends, and difficult to support the operation and maintenance decision-making of wind turbine blades throughout their entire life cycle.

[0008] Therefore, there is an urgent need for an intelligent inspection solution for wind turbine blades that is adaptable to complex working conditions, has high detection accuracy, intelligent control, reliable protection, and sufficient data application depth. Summary of the Invention

[0009] The purpose of this invention is to propose an intelligent inspection robot for wind turbine blades and its control method, which overcomes the defects of existing technologies. Through hardware scenario adaptation, lightweight algorithm optimization, closed-loop control design, hierarchical protection construction, and data twin fusion, it achieves high precision, high efficiency, high safety, and intelligent management of the entire life cycle of wind turbine blade inspection.

[0010] To achieve the above objectives, the present invention provides an intelligent inspection robot for wind turbine blades, comprising: The main hardware component, equipped with a moving mechanism, safety protection components, and a positioning module, is used to move inside the wind turbine blade cavity and ensure operational safety. The multi-view inspection module integrates a visible light camera with an adaptive dust removal unit, an infrared thermal imaging sensor, and a supplementary lighting component to collect multi-dimensional image data of the inside of the leaf cavity. The gimbal assembly is used to adjust the shooting angle of the multi-view inspection module; The central control unit is electrically connected to the main hardware, the multi-view inspection module, and the gimbal assembly, and supports bidirectional communication with the mobile operation platform and the back-end management system. The power supply module is electrically connected to the main hardware, multi-view inspection module, gimbal assembly, and central control unit, providing power support for all components of the robot.

[0011] Preferably, the main hardware body has an IP54 protection rating, dimensions of 430±10mm (length) × 220±10mm (width) × 230±10mm (height), and a weight of ≤8kg; the mobile mechanism adopts an independent four-wheel drive system, supports turning on the spot, has a maximum inspection movement speed of 0.5m / s, a maximum climbing angle of 30°, a maximum obstacle crossing height of 4cm, is equipped with anti-collision protective contact edges and side guide wheels, and has a blade internal positioning accuracy of ≤±10cm.

[0012] Preferably, the safety protection component includes a safety rope component, which weighs ≤3kg and has tension detection and automatic rope retraction functions. When the robot tilt angle exceeds a preset threshold, communication is interrupted, or a risk of loss of control is triggered, the automatic rope retraction action is initiated.

[0013] Preferably, the multi-view inspection module includes several visible light cameras with a resolution of ≥3840×2160 and a pixel count of ≥8 million, enabling multi-view acquisition; the adaptive dust removal unit is an ultrasonic dust removal module on the outside of the lens; the supplementary lighting component supports four levels of intensity adjustment, and the infrared thermal imaging sensor has a temperature measurement range of -20℃ to 150℃.

[0014] Preferably, the gimbal assembly has a horizontal movement range of 0°~360° and a vertical movement range of -90°~+90°, and is equipped with an independent supplementary lighting module that works in conjunction with the supplementary lighting component of the multi-angle inspection module to ensure sufficient lighting for shooting.

[0015] A control method for an intelligent inspection robot for wind turbine blades includes the following steps: Step S1: System configuration and task distribution. The backend management system creates hierarchical accounts and configures corresponding permissions, adds wind turbine information and builds a digital twin base for the blades, sets initial inspection parameters on the mobile operation platform and distributes them to the robot, and the robot enters the standby state after completing the equipment self-inspection. Step S2: Adaptive closed-loop inspection execution. The robot enters the blade cavity according to the initial parameters and moves forward. It collects blade cavity working condition data and imaging data in real time. Based on the working condition and defect identification results, it dynamically adjusts the inspection parameters and moving status, and simultaneously realizes the adaptive switching of the inspection mode. Step S3: Multimodal defect identification and precise localization. The imaging data is calibrated for distortion and deduplication by using a lightweight multimodal defect identification algorithm. Defects inside the leaf cavity are automatically identified. The precise location of the defect in the digital twin model of the leaf is determined by combining multi-source fusion location data. Step S4: Twin fusion data processing and uploading. The robot stores the inspection data in the vehicle storage device and synchronizes it to the back-end management system through the one-click upload function. The system completes the registration and fusion of the inspection data and the blade digital twin model, panoramic stitching and full life cycle statistical analysis of defects. Step S5: Report generation and graded emergency handling. The background management system automatically generates a standardized electronic inspection report. When safety protection conditions are triggered during the inspection, the robot performs corresponding level of stopping, alarm, posture adjustment or emergency recovery actions.

[0016] Preferably, in step S2, the inspection mode includes a semi-automatic inspection mode and a manual inspection mode. The system automatically switches modes according to real-time operating conditions. In the semi-automatic inspection mode, the wind turbine and blade information, inspection depth, travel speed, number of consecutive photos, interval photo distance, and supplementary light intensity parameters can be preset. During the inspection, the operator can manually access the control. In the manual inspection mode, the robot can be directly remotely controlled through the mobile operating platform to complete the entire inspection operation.

[0017] Preferably, in step S3, the defects include four types: bulges, wrinkles, delamination, and cracks. The lightweight multimodal defect recognition algorithm is a YOLOv8 optimized model with channel pruning and quantization compression, which integrates the prior features of the blade structure, visible light texture features, and infrared temperature features. The recognition accuracy of the four types of defects is ≥99%. The defect location needs to be accurately marked in the infrared image, visible light image, and blade digital twin model.

[0018] Preferably, in step S4, the background management system uses a 360° image fusion algorithm to register the front-view, left front-view, right front-view, and gimbal main view images with the blade digital twin model to synthesize a panoramic image of the current position. Then, it uses an image stitching algorithm to generate a panoramic twin model of the entire blade cavity, which supports statistical data on defect type and regional distribution, and analyzes the defect development trend through a fatigue damage accumulation model in combination with wind turbine operating load data.

[0019] Preferably, in step S5, the safety protection conditions include the robot triggering the anti-collision protection edge, the tilt angle exceeding the preset threshold, communication abnormality, or the battery level being lower than the warning value. The emergency handling is divided into three levels: Level 1 alarm executes stopping and attitude fine-tuning to avoid obstacles; Level 2 alarm executes slow rope retraction and communication reconnection; Level 3 alarm executes emergency rope retraction and equipment recovery. At the same time, the corresponding level of alarm prompt is issued simultaneously on the mobile operation platform.

[0020] Therefore, the present invention employs the above-described intelligent inspection robot for wind turbine blades and its control method, which has the following advantages: (1) Hardware scene adaptation, imaging quality and environmental adaptability are greatly improved: The robot adopts a compact and lightweight design, and its size and weight are strictly adapted to the narrow space of the leaf cavity. The IP54 protection level meets the needs of harsh working conditions. The 8-megapixel camera realizes blind-spot-free full-view acquisition. The ultrasonic dust removal module solves the problem of image blurring caused by dust adhesion. The four-level adjustable supplementary light and the gimbal linkage supplementary light work together to ensure the image clarity under different lighting conditions. (2) Lightweight algorithm and multimodal fusion achieve a qualitative breakthrough in defect recognition accuracy: The YOLOv8 lightweight model optimized for low computing power robot hardware balances real-time detection and accuracy; the multimodal recognition that integrates prior features of blade structure, visible light texture features and infrared temperature features can accurately identify hidden defects under the coating. The recognition accuracy of four core defects, namely bulges, wrinkles, delamination and cracks, is ≥99%, which is far higher than the existing level in the industry. (3) Adaptive closed-loop inspection control achieves a perfect balance between efficiency and accuracy: abandoning the rigid logic of manually switching inspection modes, the system can dynamically adjust the inspection parameters and travel status according to the real-time working conditions and defect identification results, intelligently switch inspection modes, efficiently inspect defect-free areas, and finely detect suspected defect areas. The inspection efficiency is more than 50% higher than the traditional fixed parameter scheme, while completely solving the problem of missed defect detection. (4) Graded adaptive safety protection, and comprehensive upgrade of operational safety and reliability: a three-level graded emergency response mechanism is constructed to perform differentiated emergency actions for different risk levels, which not only avoids inspection stoppage caused by minor contact, but also eliminates equipment fall accidents under high-risk working conditions; when communication is interrupted, local autonomous inspection and breakpoint resume can be realized, which completely solves the problem of data loss and task interruption caused by communication interruption. (5) Deep application of data driven by digital twin to realize the operation and maintenance of blades throughout the entire life cycle: Based on the digital twin model of blades, the inspection data and blade structure are accurately registered and integrated to generate a full-blade panoramic twin model; combined with the wind turbine operating load data, the defect development trend is quantitatively analyzed through the fatigue damage accumulation model, realizing the leap from "post-event detection" to "pre-event warning", providing accurate and comprehensive data support for wind turbine blade operation and maintenance decisions.

[0021] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the overall structure of the intelligent inspection robot for wind turbine blade cavity in an embodiment of the present invention. Figure 2 This is a flowchart of the method steps in an intelligent inspection robot for wind turbine blades and its control method according to an embodiment of the present invention.

[0023] Figure label: 1. Main hardware unit; 2. Multi-view inspection module; 3. Gimbal assembly; 4. Safety protection assembly; 5. Central control unit; 6. Power supply module; 7. Anti-collision protection edge; 8. Side guide wheel; 9. Positioning module; 10. Fill light assembly; 11. Adaptive dust removal unit; 12. Independent fill light module. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0025] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0026] Example like Figure 1-2 As shown in the figure, this embodiment proposes an intelligent inspection robot for wind turbine blades, specifically including: (1) Hardware Main Unit 1: The main hardware unit 1 uses a high-strength, flame-retardant ABS engineering plastic unibody shell, which is double-sealed to achieve an IP54 protection rating. It can operate stably in harsh environments with temperatures ranging from -20℃ to 55℃ and relative humidity from 5% to 95% (non-condensing). The unit's dimensions are precisely controlled at 430mm × 220mm × 230mm, and the total weight is 7.6kg. It can adapt to the narrow space requirements of mainstream megawatt-level wind turbine blade cavities (minimum inner diameter ≥ 500mm).

[0027] The mobile mechanism adopts an independent four-wheel drive design, with each wheel equipped with a 50W DC geared motor and a transmission ratio of 1:30. It supports turning on the spot (minimum turning radius of 0) and is adaptable to the curved path of the blade cavity. The maximum moving speed is 0.5m / s, and the climbing angle under rated load is 30°, which can easily cross obstacles such as 4cm high welds and protrusions inside the blade cavity. Anti-collision protective contact edges 7 (effective contact length 380mm) are installed at both ends of the body, made of elastic rubber, with a trigger pressure threshold of 5N. Two sets of side guide wheels 8 (diameter 30mm) are installed on each side of the body to reduce friction damage between the body and the inner wall of the blade cavity, and at the same time assist in the correction of the body posture.

[0028] The positioning module 9 integrates a MEMS inertial navigation unit (IMU), a wheeled odometer, and a visual feature matching unit. It uses a Kalman filter algorithm to fuse multi-source data and can still achieve a positioning accuracy of ≤±10cm in the metal shielding environment of the blade cavity, providing accurate coordinate support for defect location.

[0029] The core of safety protection component 4 is the safety rope assembly, weighing 2.7kg. It includes a high-strength nylon safety rope (5mm diameter, breaking strength ≥5kN), a high-precision tension sensor (range 0~500N, accuracy ±1N), and an adaptive rope-retracting motor (rated speed 300rpm, rope-retracting speed 0.05~0.3m / s steplessly adjustable). The safety rope assembly communicates with the central control unit 5 in real time. When the robot's tilt angle exceeds a preset threshold, communication is interrupted for more than 30 seconds, or a risk of loss of control is triggered, the automatic rope-retracting action is immediately initiated.

[0030] (2) Multi-view inspection module 2: The multi-view inspection module 2 integrates five Sony IMX586 visible light cameras, each with a resolution of 3840×2160 (8 million pixels) and a frame rate of 30fps. These cameras are installed at the front, left front, right front, front of the gimbal, and rear of the camera body, respectively, to achieve full-view coverage of the front, left front, right front, gimbal main view, and rear view, with no blind spots.

[0031] Each visible light camera lens is equipped with an adaptive dust removal unit 11, specifically a miniature ultrasonic dust removal module with an operating frequency of 40kHz and a rated power of 5W. This unit is activated when the lens dust detection sensor detects a dust concentration ≥0.5mg / m³. 3 When activated, ultrasonic vibration dust removal is automatically initiated, which, combined with the lens's hydrophobic and oleophobic coating, completely resolves the imaging blurring problem caused by dust adhesion in the leaf cavity.

[0032] The supplementary lighting assembly 10 provides each camera with an independent LED supplementary light, supporting four intensity levels: 100lm, 300lm, 500lm, and 800lm, which can be manually / automatically adjusted according to the ambient light intensity. The infrared thermal imaging sensor uses the FLIRBoson core, with a pixel size of 640×512, a temperature measurement range of -20℃ to 150℃, and a temperature measurement accuracy of ±0.5℃. It can capture temperature changes of 0.1℃ on the blade surface, helping to identify hidden defects such as delamination and cracks under the coating.

[0033] (3) Gimbal Component 3: The gimbal assembly 3 is driven by a high-precision dual-axis stepper motor, with a horizontal rotation range of 0°~360°, an angular velocity of 10° / s, and a positioning accuracy of ±0.1°; the vertical pitch range is -90°~+90°, enabling precise focusing at any angle inside the leaf cavity. The gimbal is equipped with an independent supplementary lighting module 12, with a power of 10W and a color temperature of 5500K. It works in conjunction with the supplementary lighting assembly 10 of the multi-view inspection module 2. When the gimbal rotates, the independent supplementary lighting module 12 synchronously adjusts the supplementary lighting angle, with a supplementary lighting angle deviation of ≤5°, ensuring uniform and sufficient light when shooting from complex angles.

[0034] (4) Central control unit 5: The central control unit 5 uses an NVIDIA Jetson Xavier NX embedded processor with a main frequency of 1.9GHz and integrated CUDA cores to support AI algorithm acceleration processing; it has built-in 16GB eMMC storage, expandable to 128GB, for local storage of inspection data. The communication module integrates 2.4G / 5G dual-band WiFi and a LoRa gateway, with a line-of-sight communication distance of ≥200m, meeting the requirements for stable bidirectional data transmission in the metal-shielded environment of the leaf cavity. It can receive control commands from the mobile operation platform in real time and simultaneously upload inspection data to the background management system.

[0035] The central control unit 5 has a built-in lightweight multimodal defect recognition engine, which can realize local real-time processing of image data. It also has core control logic for working condition perception, dynamic parameter adjustment, and hierarchical emergency handling, and is the "brain" of the robot.

[0036] (5) Power supply module 6: The power supply module 6 uses a 24V / 15Ah high-rate lithium polymer battery pack with an energy density of 210Wh / kg. Under full load (with all cameras, gimbals, and communication modules activated), it provides a runtime of ≥4 hours and supports full-range single-blade inspection (single inspection distance ≤1.5km). The battery pack has a built-in BMS battery management system with overvoltage, undervoltage, overcurrent, and overtemperature protection functions. It supports fast charging (fully charged in 2 hours) and can simultaneously provide real-time feedback of remaining power, voltage, temperature, and other status data to the central control unit 5.

[0037] The power supply module 6 is electrically connected to the main hardware 1, the multi-view inspection module 2, the gimbal assembly 3, and the central control unit 5 via an aviation plug, providing a stable DC power supply to each component.

[0038] A control method for an intelligent inspection robot for wind turbine blades, comprising the following steps: Step S1, System Configuration and Task Distribution Stage: The backend management system is deployed on an industrial-grade intelligent server, with a minimum configuration of: a 12-core, 24-thread 2.1GHz CPU, two 32GB DDR4 ECC 3200MHz memory modules, three 4TB SATA hard drives, a 2GB cache RAID card, dual 900W redundant power supplies, dual-port gigabit network cards, and an RTX 3080 graphics card. The system supports integrated management of multiple sites and allows for the creation of three-tiered accounts: super administrator, regional administrator, and site administrator. Differentiated operating permissions can be configured for each account to ensure data security and hierarchical management.

[0039] Operators add information about wind farm stations and wind turbines through the back-end management system, including basic parameters such as wind turbine model, blade length, 3D model of blade cavity, ply structure, and installation location, to build a digital twin base for wind turbine blades, which serves as the digital foundation for subsequent inspections, defect location, and data fusion.

[0040] The mobile operating platform uses a 10-inch industrial-grade touchscreen tablet with a battery life of ≥6 hours. After logging into their authorized account, operators select the target wind turbine and blades, set initial inspection parameters (inspection depth, baseline travel speed, photo interval, supplementary lighting level, etc.), and send them to the inspection robot. After receiving the instructions, the robot completes a full self-check of battery power, camera status, communication connection, safety rope tension, and gimbal function. If no abnormalities are found during the self-check, it enters the standby state.

[0041] Step S2, Adaptive Closed-Loop Inspection Execution Phase: The operator inserts the robot into the blade cavity through the mounting interface at the root of the wind turbine blade and starts the inspection task. The robot moves along the central axis of the blade towards the blade tip according to the initial parameters. During the movement, it collects real-time operating data of the blade cavity (body attitude, ambient light intensity, communication quality, obstacle distribution) and imaging data, and simultaneously performs real-time preliminary defect judgment through the lightweight multimodal defect recognition engine of the central control unit 5.

[0042] Based on real-time operating conditions and initial defect judgment results, the system dynamically adjusts inspection parameters and travel status: in flat areas without defects or obstacles, the travel speed is automatically increased to 0.5m / s, the photo interval is expanded, and the inspection efficiency is improved; when suspected defects, obstacles, or entering the blade bending area are identified, the travel speed is automatically reduced to 0.1m / s, the photo interval is reduced, and refined inspection is initiated.

[0043] Meanwhile, the system achieves intelligent adaptive switching between inspection modes: under normal operating conditions, it maintains a semi-automatic inspection mode, with the robot autonomously completing inspection tasks; when path deviation, suspected defects, or communication quality degradation are detected, it automatically switches to a semi-manual mode, pausing its movement and prompting the operator to manually intervene for confirmation; in extreme conditions such as communication interruption or abnormal robot posture, it switches to a manual remote control mode, allowing the operator to remotely control the robot's operation. During semi-automatic inspection, the operator can manually access the control at any time to adjust the robot's movement status, gimbal angle, and camera capture actions, achieving refined inspection.

[0044] Step S3, Multimodal Defect Identification and Precise Localization Stage: The central control unit 5 runs a lightweight multimodal defect recognition algorithm to process the acquired image data throughout the entire process. (1) Image preprocessing: Zhang's calibration method was used to calibrate the visible light and infrared images for distortion. The radial distortion correction coefficients k1=-0.04 and k2=0.002, and the tangential distortion correction coefficients p1=0.001 and p2=-0.0005 were used to eliminate the image deformation caused by camera optical distortion and leaf cavity surface reflection. (2) Defect identification: The YOLOv8 optimized model with channel pruning and quantization compression is adopted. The pruning rate is 30%, the number of model parameters is reduced from 25.9M to 18.1M, and the inference speed is improved by 40%, which is suitable for the real-time detection requirements of low computing power robot hardware. The algorithm integrates the prior features of blade structure, visible light texture features and infrared temperature features to automatically identify four types of core defects: bulge, wrinkle, delamination and crack. The identification accuracy is ≥99%. (3) Defect deduplication: The ORB feature matching algorithm is used to compare the defect regions in adjacent images. If the feature similarity is ≥85%, they are determined to be the same defect and deduplication is automatically performed to ensure that the defects are globally unique. (4) Precise positioning: Combining the three-dimensional coordinate data (X: blade axial distance, Y: circumferential angle, Z: radial depth) of the positioning module 9, the defect boundary is marked in the visible light image and infrared image, and the defect coordinates are precisely mapped to the blade digital twin model to realize the three-dimensional visualization annotation of the defect location with a positioning error ≤ ±10cm.

[0045] Step S4, Twin Fusion Data Processing and Upload Stage: After completing the inspection task, the robot returns to the blade root and stores all inspection data (image data, defect information, machine status log, environmental parameters, etc.) to an onboard USB flash drive. The operator then inserts the USB flash drive into the backend management system server and uses the one-click upload function to synchronize the data. The system automatically parses the inspection data on the USB flash drive, matching it with information such as the inspection task name, time, and turbine number to automatically classify, organize, and archive the data.

[0046] The system uses a 360° image fusion algorithm to perform pixel-level registration of the front-view, left front-view, right front-view, and gimbal front-view images with the blade digital twin model, synthesizing them into a 360° panoramic image of the blade cavity at the current position, with a stitching error of ≤2 pixels. Then, using the SIFT feature matching algorithm, the panoramic images at each position are stitched sequentially along the blade axis to generate a full-blade panoramic twin model of the blade cavity, completely presenting the internal structure and defect distribution of the entire blade length.

[0047] The system supports multi-dimensional defect statistical analysis: it counts the number and proportion of defects by type, and analyzes defect distribution by blade root, middle, and tip regions. By comparing historical inspection data, it generates defect development trend curves. Simultaneously, it combines real-time wind turbine operating load data (wind speed, rotational speed, blade sway angle, vibration data) and uses the Miner fatigue damage accumulation model to calculate the fatigue propagation rate of defects, predict the time when defects reach critical failure size, and generate tiered maintenance early warning suggestions, achieving intelligent maintenance management throughout the blade's entire lifecycle. The system supports manual review and modification of defect information; operators can correct defect types, supplement defect descriptions, and delete misjudged defects to ensure data accuracy.

[0048] Step S5, Report Generation and Tiered Emergency Response Phase: The back-end management system automatically generates standardized electronic inspection reports based on the processed inspection data. The report includes: basic information of the wind farm station and wind turbine, inspection time and environmental parameters, defect details (type, precise location, size, and visible light / infrared image evidence), defect statistical distribution charts, defect development trend analysis and operation and maintenance suggestions. The report supports one-click export and printing in PDF format.

[0049] During the inspection, the system monitors the robot's operating status in real time. When safety protection conditions are triggered, it executes a three-level emergency response and simultaneously issues corresponding audible and visual alarms on the mobile operating platform. Level 1 alarm: The triggering conditions are slight contact of the anti-collision protection edge 7 (pressure 3N≤P<5N) and slight tilt of the machine body (left and right tilt angle 10°≤θ<15°). Emergency action: stop immediately, and use the side guide wheel 8 and drive wheel to finely adjust the body posture to avoid obstacles. After the posture returns to normal, the inspection can continue to avoid inspection stoppage caused by slight contact. Level 2 alarm: Triggering conditions are: anti-collision protection contact edge 7 fully triggered (pressure ≥ 5N), body tilt exceeding the standard (15° ≤ θ < 25°), communication interruption 10s ≤ t < 30s, and battery level below 20%. Emergency action: stop the machine immediately, start the safety rope and slowly reel it in (speed 0.1m / s), while continuously trying to reconnect the communication. After the communication is restored, the operator should confirm whether to continue the inspection. Level 3 alarm: Triggering conditions are severe tilt of the robot body (θ≥25°), communication interruption ≥30s, battery level below 5%, and abnormal tension detected by the tension sensor ≥300N. Emergency action: Immediately cut off the power to the drive motor, start the emergency rope retraction (speed 0.3m / s), and quickly retrieve the robot to the safe area at the root of the blade. The mobile operation platform will continue to issue alarms until the operator confirms the handling.

[0050] All emergency response actions were fully recorded in the aircraft status log for easy tracing and analysis later.

[0051] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A smart inspection robot for wind turbine blades, characterized in that, include: The main hardware component, equipped with a moving mechanism, safety protection components, and a positioning module, is used to move inside the wind turbine blade cavity and ensure operational safety. The multi-view inspection module integrates a visible light camera with an adaptive dust removal unit, an infrared thermal imaging sensor, and a supplementary lighting component to collect multi-dimensional image data of the inside of the leaf cavity. The gimbal assembly is used to adjust the shooting angle of the multi-view inspection module; The central control unit is electrically connected to the main hardware, the multi-view inspection module, and the gimbal assembly, and supports bidirectional communication with the mobile operation platform and the back-end management system. The power supply module is electrically connected to the main hardware, multi-view inspection module, gimbal assembly, and central control unit, providing power support for all components of the robot.

2. The intelligent inspection robot for wind turbine blades according to claim 1, characterized in that: The main hardware unit has an IP54 protection rating, dimensions of 430±10mm (length) × 220±10mm (width) × 230±10mm (height), and a weight of ≤8kg. The mobile mechanism adopts an independent four-wheel drive system, supports turning on the spot, has a maximum inspection movement speed of 0.5m / s, a maximum climbing angle of 30°, a maximum obstacle crossing height of 4cm, and is equipped with anti-collision protective contact edges and side guide wheels. The positioning accuracy within the blades is ≤±10cm.

3. The intelligent inspection robot for wind turbine blades according to claim 1, characterized in that: The safety protection components include a safety rope assembly, which weighs ≤3kg and has tension detection and automatic rope retraction functions. When the robot tilt angle exceeds a preset threshold, communication is interrupted, or a risk of loss of control is triggered, the automatic rope retraction action is initiated.

4. The intelligent inspection robot for wind turbine blades according to claim 1, characterized in that: The multi-view inspection module includes several visible light cameras with a resolution of ≥3840×2160 and a pixel count of ≥8 million, enabling multi-view acquisition; the adaptive dust removal unit is an ultrasonic dust removal module on the outside of the lens; the supplementary lighting component supports four levels of intensity adjustment, and the infrared thermal imaging sensor has a temperature measurement range of -20℃ to 150℃.

5. The intelligent inspection robot for wind turbine blades according to claim 1, characterized in that: The gimbal assembly has a horizontal movement range of 0°~360° and a vertical movement range of -90°~+90°. It is equipped with an independent fill light module, which works in conjunction with the fill light component of the multi-angle inspection module to ensure sufficient lighting for shooting.

6. A control method for an intelligent inspection robot for wind turbine blades, applied to the intelligent inspection robot for wind turbine blades as described in any one of claims 1-5, characterized in that, Includes the following steps: Step S1: System configuration and task distribution. The backend management system creates hierarchical accounts and configures corresponding permissions, adds wind turbine information and builds a digital twin base for the blades, sets initial inspection parameters on the mobile operation platform and distributes them to the robot, and the robot enters the standby state after completing the equipment self-inspection. Step S2: Adaptive closed-loop inspection execution. The robot enters the blade cavity according to the initial parameters and moves forward. It collects blade cavity working condition data and imaging data in real time. Based on the working condition and defect identification results, it dynamically adjusts the inspection parameters and moving status, and simultaneously realizes the adaptive switching of the inspection mode. Step S3: Multimodal defect identification and precise localization. The imaging data is calibrated for distortion and deduplication by using a lightweight multimodal defect identification algorithm. Defects inside the leaf cavity are automatically identified. The precise location of the defect in the digital twin model of the leaf is determined by combining multi-source fusion location data. Step S4: Twin fusion data processing and uploading. The robot stores the inspection data in the vehicle storage device and synchronizes it to the back-end management system through the one-click upload function. The system completes the registration and fusion of the inspection data and the blade digital twin model, panoramic stitching and full life cycle statistical analysis of defects. Step S5: Report generation and graded emergency handling. The background management system automatically generates a standardized electronic inspection report. When safety protection conditions are triggered during the inspection, the robot performs corresponding level of stopping, alarm, posture adjustment or emergency recovery actions.

7. The control method for an intelligent inspection robot for wind turbine blades according to claim 6, characterized in that: In step S2, the inspection mode includes a semi-automatic inspection mode and a manual inspection mode. The system automatically switches modes according to real-time operating conditions. In the semi-automatic inspection mode, the fan and blade information, inspection depth, travel speed, number of consecutive photos, interval photo distance and supplementary light intensity parameters can be preset. During the inspection, the operator can manually access the control. In manual inspection mode, the robot can be remotely controlled directly through a mobile operating platform to complete the entire inspection process.

8. The control method for an intelligent inspection robot for wind turbine blades according to claim 6, characterized in that: In step S3, the defects include four types: bulges, wrinkles, delamination, and cracks. The lightweight multimodal defect recognition algorithm is a YOLOv8 optimized model with channel pruning and quantization compression, which integrates the prior features of the blade structure, visible light texture features, and infrared temperature features. The recognition accuracy of the four types of defects is ≥99%. The defect location needs to be accurately marked in the infrared image, visible light image, and blade digital twin model at the same time.

9. The control method for an intelligent inspection robot for wind turbine blades according to claim 6, characterized in that: In step S4, the background management system uses a 360° image fusion algorithm to register the front-view, left front-view, right front-view, and gimbal main view images with the blade digital twin model, synthesizing them into a panoramic view of the current position. Then, it uses an image stitching algorithm to generate a panoramic twin model of the entire blade cavity, supporting statistical data on defect type and regional distribution. Combined with wind turbine operating load data, it analyzes the defect development trend through a fatigue damage accumulation model.

10. The control method for an intelligent inspection robot for wind turbine blades according to claim 6, characterized in that: In step S5, safety protection conditions include the robot triggering the anti-collision protection edge, tilt angle exceeding the preset threshold, communication abnormality, or battery level below the warning value. Emergency handling is divided into three levels: Level 1 alarm executes stopping and attitude fine-tuning to avoid obstacles; Level 2 alarm executes slow rope retraction and communication reconnection; Level 3 alarm executes emergency rope retraction and equipment recovery. At the same time, alarm prompts of the corresponding level are issued simultaneously on the mobile operation platform.

Images