A wind turbine blade inspection system and method based on unmanned aerial vehicles (UAVs)

Abstract

This invention discloses a wind turbine blade inspection system and method based on unmanned aerial vehicles (UAVs). The invention relates to the field of wind turbine blade inspection technology and includes the following steps: a priority analysis module collects historical inspection and operational data of each wind turbine blade in a wind farm, constructs a health scoring model, and generates a health priority queue; an inspection target determination module selects wind turbines to be inspected based on the initial flight distance of the UAV and the health priority queue, forming a target set; an inspection route planning module plans an inspection path based on the target set and the spatial distribution of the wind turbines; a blade defect detection module controls the UAV to inspect along the planned path, collects image data, and identifies defects; a real-time target adjustment module analyzes wind turbines with defects, determines the associated detection areas, adds uninspected wind turbines to the target set, and regenerates the inspection path to ensure the completion of all inspection tasks, improving the flexibility and resource utilization efficiency of the inspection tasks.

Description

Technical Field

[0001] This invention relates to the field of wind turbine blade inspection technology, specifically to a wind turbine blade inspection system and method based on unmanned aerial vehicles (UAVs). Background Technology

[0002] Offshore wind turbine blades are a crucial component of wind turbines, and their operational status directly impacts the power generation efficiency and safety of the wind turbine unit. Because offshore wind farms are constantly exposed to the marine environment, the blades are subject to erosion from sea winds, high humidity, salt spray, ultraviolet radiation, and waves, making them more prone to various defects than onshore wind turbines. These defects include surface corrosion, fouling, and structural damage such as cracks. These defects can lead to decreased aerodynamic performance, reduced turbine operating efficiency, and even serious blade breakage and complete turbine shutdown, resulting in significant economic losses for wind farm operations.

[0003] Current drone inspection tasks are typically based on fixed inspection cycles or wind turbine locations, failing to prioritize the assessment of wind turbine blade health by combining historical inspection data with real-time operational data. This approach may lead to inefficient resource allocation and make it difficult to conduct focused inspections of high-risk wind turbines.

[0004] In the prior art, CN117536797A discloses a wind turbine blade inspection system and method based on a drone. This system includes: a data acquisition module for acquiring blade data and surrounding environmental data of the blade to be inspected, determining characteristic inspection points of the drone based on the blade data and surrounding environmental data, and generating an inspection path; an inspection module for taking pictures of the corresponding blades at the characteristic inspection points according to the inspection path, obtaining a first inspection image and a second inspection image; and an analysis module for receiving the first and second inspection images, analyzing them to obtain the health status of the corresponding blades, and formulating a maintenance plan based on the blade health status. However, this system ignores the priority relationship between inspected wind turbines, and cannot adjust subsequent inspection plans based on real-time detection results during the inspection task execution. For example, if a blade has a serious defect, it may indicate that adjacent wind turbine blades also have similar problems, but this method cannot dynamically adjust the inspection priority and path, thus affecting the inspection efficiency and effectiveness.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a wind turbine blade inspection system and method based on unmanned aerial vehicles (UAVs) to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A wind turbine blade inspection system based on unmanned aerial vehicles (UAVs) specifically includes:

[0009] The priority analysis module is used to collect historical inspection data of each wind turbine blade in the wind farm and working data for a specified period before shutdown. After normalizing the data, a health scoring model is constructed to score the health status of each blade and generate a wind turbine health priority queue sorted by health priority.

[0010] The inspection target determination module is used to select inspection wind turbines from the queue to form an inspection target set based on the initial flight range of the UAV and the spacing between adjacent wind turbines in the health priority queue.

[0011] The inspection route planning module is used to plan the inspection path of the UAV based on the inspection target set and using the nearest neighbor heuristic algorithm.

[0012] The blade defect detection module is used to control the drone to inspect the target wind turbine sequentially according to the planned inspection path, and to collect image data of the blades during the inspection process. Based on the changes in grayscale values ​​in the image, the module identifies the target area, extracts the geometric morphology parameters of the target area, and determines the type of blade defect by combining the vibration characteristic data of the corresponding blade.

[0013] The real-time target adjustment module is used to analyze the degree of structural defects in blades with structural defects, determine the associated detection areas, add the uninspected wind turbines in the associated detection areas to the target set of this round of inspection, and regenerate the inspection path for the remaining wind turbines to be inspected in this round of inspection in order to complete all inspection tasks.

[0014] Furthermore, the historical inspection data includes the number of historical failures and the number of years of operation for each blade, and the working data for the specified period before shutdown includes differences in blade vibration amplitude, blade rotation speed, and power generation.

[0015] Specifically, the differences in vibration amplitude, blade rotation speed, and power generation refer to the absolute differences between the blade vibration amplitude, rotation speed, and power generation within a specified time period before shutdown and the reference vibration amplitude, rotation speed, and power generation at the same wind speed.

[0016] A health scoring model is constructed to score the health status of each leaf. The formula used to calculate the leaf health status score is as follows:

[0017] ;

[0018] In the formula, Score the health status of the i-th leaf. Let be the normalized value of the historical failure count for the i-th blade. Let be the normalized value of the service life of the i-th blade. Let be the state coefficient of the i-th leaf. and These are the historical scaling constant and the actual scaling constant, respectively, where i is the index of the wind turbine blade;

[0019] Where the state coefficient of the i-th blade Specifically, it is characterized by the working data of a specified period before shutdown, and the specific formula used for calculation is as follows:

[0020] ;

[0021] In the formula, Let be the normalized value of the vibration amplitude difference of the i-th blade. Let be the normalized value of the rotational speed difference for the i-th blade. Let be the normalized value of the power generation difference of the i-th blade.

[0022] Furthermore, the average health status score of the blades in each wind turbine is calculated, and the wind turbines are ranked according to the average health status score to form a wind turbine health priority queue. The formula used to calculate the average health status score of each wind turbine is as follows:

[0023] ;

[0024] In the formula, Let be the average health status score corresponding to the j-th wind turbine. For the j-th wind turbine, the health status score of the i-th blade is given. denoted by , where is the number of blades in the wind turbine, and j is the index of the wind turbine within the wind farm;

[0025] Based on the average health status score, the wind turbines in the wind farm are ranked in descending order to form a wind turbine health priority queue.

[0026] Furthermore, the logic for constructing the inspection target set is as follows: set the take-off and landing positions of the UAV, with the constraint that the length of the preset flight path is not higher than the initial endurance distance, add wind turbines in sequence according to the wind turbine health priority queue to update the preset flight path, and take each wind turbine in the final preset flight path as the inspection wind turbine to form the inspection target set, and take the wind turbines in the wind farm that have not been added to the preset flight path as candidate inspection wind turbines. The preset flight path is the path that passes through each inspection wind turbine in sequence according to the order of wind turbine addition, with the take-off position as the starting point and the landing position as the ending point. The initial endurance distance specifically refers to the maximum flight distance of the UAV when the battery is not consumed.

[0027] Furthermore, the inspection path is the shortest flight route covering each wind turbine in the inspection target set. The specific logic of using the nearest neighbor heuristic algorithm to plan the inspection path of the UAV is as follows: taking the take-off position as the mission start point and the landing position as the mission end point, taking each wind turbine in the inspection target set as the path target point, and determining the shortest flight route covering each wind turbine in the inspection target set as the inspection path.

[0028] Furthermore, the types of blade defects include structural defects and stain defects;

[0029] The steps to determine the target area are as follows: perform background separation processing on the collected leaf image data, convert the leaf part in the image into a grayscale image, extract and analyze the grayscale value of each pixel in the grayscale image, and determine the target area based on the gradient change of the grayscale value of the pixel.

[0030] The logic for determining the target region based on the gradient change of pixel grayscale values ​​is as follows: the gradient of each pixel in the blade image is calculated using the Sobel operator, a gradient change threshold is set, pixels with gradients greater than the gradient change threshold are taken as boundary points, and the closed region formed by the boundary points is taken as the target region; the vibration feature data specifically refers to the blade vibration amplitude and vibration dominant frequency value.

[0031] The logic behind determining the type of blade defect by combining vibration characteristic data of the corresponding blade within a specified time period before shutdown is as follows: The blade vibration amplitude data is extracted from the blade's vibration time history data, and the dominant vibration frequency is analyzed. Based on the degree of deviation of the dominant vibration frequency and the difference in blade vibration amplitude, a structural defect coefficient is calculated to determine whether a structural defect exists on the blade. The specific formula for calculating the structural defect coefficient is as follows:

[0032] ;

[0033] In the formula, Let be the structural defect coefficient of the i-th blade. Let be the dominant vibration frequency of the i-th blade. This is the vibration reference dominant frequency value for the i-th blade;

[0034] The logic used to determine whether a blade has structural defects is as follows:

[0035] like At that time, it was determined that the blade had a structural defect;

[0036] like At that time, it was determined that the blade did not have structural defects, and the target area was a stain defect; The threshold for judging structural defects;

[0037] The type of structural defect is determined based on the geometric parameters of the target area, including the area, length and circularity of the area, and the type of structural defect includes blade cracks and surface corrosion.

[0038] The specific judgment logic is as follows: if the area and roundness of the region are less than the set area threshold and roundness threshold, but the corresponding length is greater than the set length threshold, the structural defect is judged to be a blade crack; if the area and roundness of the region are greater than the set area threshold and roundness threshold, but the corresponding length is less than the set length threshold, the structural defect is judged to be surface corrosion.

[0039] Furthermore, the logic for determining the specific associated detection area is as follows: Real-time detection and output of blade defect types; if structural defects are detected in the blade, an associated detection area is defined with the current UAV position as the center. The radius of the associated detection area is determined by the defect severity, specifically by combining the crack length with the area of ​​surface corrosion. The formula for calculating the radius of the associated detection area is as follows:

[0040] ;

[0041] In the formula, The radius of the correlation detection domain, As a reference radius, This represents the total length of the blade's slits. The reference length of the crack. For length influence constant, The constant is the influence of corrosion area. This represents the total corrosion area of ​​the blade. Corrosion reference area;

[0042] Candidate wind turbines in the correlation detection field are added to the inspection target set. Taking the current position of the UAV as the starting point and the landing position as the mission endpoint, the other wind turbines in the inspection target set except for those that have been inspected are taken as path target points. Candidate shortest flight routes are generated based on the nearest neighbor heuristic algorithm, and the candidate shortest flight routes are compared with the current flight range of the UAV.

[0043] If the current flight range of the drone is less than the candidate shortest flight path, the corresponding wind turbine blade inspection will be completed according to the initial inspection path, and the candidate wind turbines in the associated inspection field will be used as the inspection targets for the next round.

[0044] If the drone's current range is not greater than the shortest flight path, then the inspection will be carried out according to the candidate shortest flight path.

[0045] This invention also provides a method for inspecting wind turbine blades based on unmanned aerial vehicles (UAVs). This method is used to control the aforementioned method for inspecting wind turbine blades based on UAVs, and includes:

[0046] Historical inspection data and working data for a specified period before shutdown of each wind turbine blade in the wind farm are collected. After normalizing the data, a health scoring model is constructed to score the health status of each blade and generate a wind turbine health priority queue sorted by health priority.

[0047] Based on the initial flight range of the drone and the spacing between adjacent wind turbines in the health priority queue, wind turbines to be inspected are selected from the queue to form an inspection target set;

[0048] Based on the set of inspection targets, the nearest neighbor heuristic algorithm is used to plan the inspection path of the UAV.

[0049] The drone is controlled to inspect the target wind turbines sequentially according to the planned inspection path, and image data of the blades is collected during the inspection. Based on the changes in grayscale values ​​in the images, the target area is identified, the geometric morphology parameters of the target area are extracted, and the blade defect type is determined by combining the vibration characteristic data of the corresponding blade.

[0050] Analyze the degree of structural defects in the blades with defects, determine the associated detection areas, add the uninspected wind turbines in the associated detection areas to the target set of this round of inspection, and regenerate the inspection path for the remaining wind turbines in this round of inspection to complete all inspection tasks.

[0051] Compared with the prior art, the beneficial effects of the present invention are:

[0052] A health scoring model is used to quantitatively assess the health status of wind turbine blades, prioritizing inspections of turbines with poor health, significantly improving the targeting and efficiency of inspection tasks. By integrating image grayscale differences, geometric parameters, and blade vibration characteristics, accurate identification of blade defects is achieved, demonstrating strong environmental adaptability and the ability to cope with complex conditions such as uneven lighting and salt spray interference in marine environments. During inspections, blade defect data is analyzed in real time, dynamically adjusting the priority of turbines that have not yet been inspected and replanning inspection routes to ensure that turbines with critical defects are detected and addressed in a timely manner, improving the flexibility and resource utilization efficiency of inspection tasks. Attached Figure Description

[0053] Figure 1 This is a schematic diagram of the overall device structure of the present invention;

[0054] Figure 2 This is a vibration spectrum diagram of a single blade.

[0055] Figure 3Vibration time history diagram for blades with structural defects;

[0056] Figure 4 This is a schematic diagram of the overall method flow of the present invention. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0058] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0059] Example:

[0060] Please see Figures 1-3 The present invention provides a technical solution:

[0061] A wind turbine blade inspection system based on unmanned aerial vehicles (UAVs) specifically includes:

[0062] The priority analysis module is used to collect historical inspection data and working data for a specified period before shutdown of each wind turbine blade in the wind farm. After normalizing the data, a health scoring model is constructed to score the health status of each blade and generate a wind turbine health priority queue sorted by health priority.

[0063] The historical inspection data includes the number of historical faults and the number of years of operation for each blade, and the working data for the specified period before shutdown includes differences in blade vibration amplitude, differences in blade rotation speed, and differences in power generation.

[0064] Wind farms typically use operation and maintenance management systems, such as SCADA systems (wind turbine monitoring systems), to record fault information for each blade, including the time and type of the fault. They retrieve historical fault occurrences from the database, and statistically analyze and store this information based on equipment maintenance logs and fault alarm information for subsequent analysis.

[0065] The installation time information of wind turbine equipment is usually recorded in the operation and maintenance management system. The installation time of the wind turbine is retrieved by the operation management database, and the difference between the current time and the installation time is calculated to obtain the service life of the blades.

[0066] Specifically, the differences in vibration amplitude, blade rotation speed, and power generation refer to the absolute differences between the blade vibration amplitude, rotation speed, and power generation within a specified time period before shutdown and the reference vibration amplitude, rotation speed, and power generation at the same wind speed.

[0067] The vibration amplitude of the blades is collected in real time by vibration sensors installed on the wind turbine blades or nacelle. Real-time vibration data of the blades is collected within a specified period, such as 30 minutes before shutdown, and stored through a SCADA system or independent data acquisition equipment.

[0068] The blade speed can be monitored in real time by the speed sensor in the wind turbine nacelle or the wind turbine control system. Blades with large differences in speed may have wear, deformation or other faults.

[0069] Power generation is the most critical operating indicator for wind turbines. It can be collected through power sensors or energy metering equipment in the wind turbine nacelle. The power sensors measure the output power of the wind turbine in real time, or through SCADA systems. The monitoring system of a wind farm usually records the power generation information of each wind turbine.

[0070] A health scoring model is constructed to score the health status of each leaf. The formula used to calculate the leaf health status score is as follows:

[0071] ;

[0072] In the formula, Score the health status of the i-th leaf. Let be the normalized value of the historical failure count for the i-th blade. Let be the normalized value of the service life of the i-th blade. Let be the state coefficient of the i-th leaf. and These are the historical scaling constant and the actual scaling constant, respectively, where i is the index of the wind turbine blade;

[0073] It should be noted that the health status score of the i-th leaf This value indicates the health of the leaves; a higher value indicates a worse leaf health condition, and vice versa.

[0074] The historical failure count and operational years reflect the failure records accumulated by the blade during its life cycle, which are usually related to blade aging and usage. Higher values ​​lead to a higher health score, indicating a decline in blade health. Therefore, the historical failure count and operational years are correlated with the health status score of the i-th blade. Proportional;

[0075] Among them, historical scaling constant and actual scaling constant and Specific settings can be configured based on expert experience.

[0076] Where the state coefficient of the i-th blade Specifically, it is characterized by the working data of a specified period before shutdown, and the specific formula used for calculation is as follows:

[0077] ;

[0078] In the formula, Let be the normalized value of the vibration amplitude difference of the i-th blade. Let be the normalized value of the rotational speed difference for the i-th blade. Let be the normalized value of the power generation difference of the i-th blade.

[0079] It should be noted that the state coefficient of the i-th blade The state coefficient is used to characterize the vibration and rotational speed differences of blades. A higher value indicates a worse blade condition; therefore, it is called the state coefficient. and It is directly proportional; the purpose of this formula is to characterize the current state of the blade through three key operating parameters of the blade: vibration, rotational speed and power generation, so as to comprehensively reflect the health and performance of the blade.

[0080] Vibration amplitude is a direct reflection of blade operational stability. Abnormal vibration may result from blade structural damage, such as cracks, loose connections, or decreased surface aerodynamic performance. Rotational speed directly determines the wind turbine's power output performance; abnormal speed may be related to uneven blade mass or decreased aerodynamic performance. Power generation is the ultimate goal of wind turbine operation, and blade condition directly affects power generation performance. In particular, when blade aerodynamic efficiency decreases, power generation will significantly decrease. Therefore, differences in vibration amplitude, rotational speed, and power generation are all related to the state coefficient. They are directly proportional. To make the differences in vibration amplitude, rotational speed, and power generation comparable, these three parameters were normalized.

[0081] Calculate the average health status score of the blades in each wind turbine, and rank the wind turbines according to the average health status score to form a wind turbine health priority queue. The formula used to calculate the average health status score of each wind turbine is as follows:

[0082] ;

[0083] In the formula, Let be the average health status score corresponding to the j-th wind turbine. For the j-th wind turbine, the health status score of the i-th blade is given. denoted by , where is the number of blades in the wind turbine, and j is the index of the wind turbine within the wind farm;

[0084] It should be noted that by calculating the average health status score of all blades in each wind turbine, the overall health status of the turbine can be effectively reflected. Assessing the health status of each blade individually may result in outliers affecting the outcome, while the average balances the impact of these extreme values. The average provides a standardized indicator, making comparisons of health status between different wind turbines more fair and reasonable. Different wind turbines may have different numbers of blades; the average eliminates the influence of these differences.

[0085] Based on the average health status score, the wind turbines in the wind farm are ranked in descending order to form a wind turbine health priority queue.

[0086] The inspection target determination module is used to select inspection wind turbines from the queue to form an inspection target set based on the initial flight range of the UAV and the spacing between adjacent wind turbines in the health priority queue.

[0087] The logic for constructing the inspection target set is as follows: set the take-off and landing positions of the UAV, with the constraint that the length of the preset flight path is not higher than the initial endurance distance, add wind turbines in sequence according to the wind turbine health priority queue to update the preset flight path, and take each wind turbine in the final preset flight path as the inspection wind turbine to form the inspection target set, and take the wind turbines in the wind farm that have not been added to the preset flight path as candidate inspection wind turbines. The preset flight path is the path that starts from the take-off position and ends at the landing position, and passes through each inspection wind turbine in the order of wind turbine addition. The initial endurance distance specifically refers to the maximum flight distance of the UAV when the battery is not consumed.

[0088] The preset flight path starts at the drone's takeoff location and ends at its landing location, ensuring that the inspection mission is completed within the drone's range. This method meets the drone's range limitations, avoids path planning exceeding the battery's maximum flight distance, and ensures mission safety. The wind turbines are prioritized based on their health status scores; turbines with higher health scores, indicating poorer health conditions, are included in the inspection path first, ensuring the rational allocation of resources and prioritizing the inspection of turbines requiring attention, thus avoiding resource waste.

[0089] The drone's flight path is constrained by its initial range. The inspection path is planned within this range to ensure mission completion while preventing inspection failure due to the drone running out of power.

[0090] The inspection route planning module is used to plan the inspection path of the UAV based on the set of inspection targets using the nearest neighbor heuristic algorithm.

[0091] The inspection path is the shortest flight route covering each wind turbine in the inspection target set. The specific logic of using the nearest neighbor heuristic algorithm to plan the inspection path of the UAV is as follows: take the take-off position as the mission start point and the landing position as the mission end point, take each wind turbine in the inspection target set as the path target point, and determine the shortest flight route covering each wind turbine in the inspection target set as the inspection path.

[0092] By using the wind turbines in the inspection target set as waypoints, all wind turbines that need to be inspected are covered. This ensures that the drone does not miss any critical targets during the inspection process;

[0093] The nearest neighbor heuristic algorithm is an efficient greedy algorithm that progressively constructs a path by selecting the nearest target point at each step, thereby minimizing the flight distance. This method is simple and effective, particularly suitable for dynamically changing inspection scenarios. By progressively selecting the nearest wind turbine, the nearest neighbor heuristic algorithm shortens the inspection path, thus improving inspection efficiency. In practice, reducing flight time saves battery power and increases inspection frequency. The nearest neighbor heuristic algorithm is simple and easy to use; compared to other more complex path optimization algorithms, its calculation process is relatively fast, and it can generate feasible paths in a short time, making it suitable for real-time inspection tasks.

[0094] In some cases, the inspection target may change, such as the temporary addition of wind turbines or sudden health problems. The nearest neighbor heuristic algorithm can quickly replan the path and flexibly adapt to new inspection requirements.

[0095] The blade defect detection module is used to control the UAV to inspect the target wind turbine sequentially according to the planned inspection path, and to collect image data of the blades during the inspection process. Based on the changes in grayscale values ​​in the image, the module identifies the target area, extracts the geometric morphological parameters of the target area, and determines the type of blade defect by combining the vibration characteristic data of the corresponding blade.

[0096] The types of blade defects include structural defects and stain defects;

[0097] The steps to determine the target area are as follows: perform background separation processing on the collected leaf image data, convert the leaf part in the image into a grayscale image, extract and analyze the grayscale value of each pixel in the grayscale image, and determine the target area based on the gradient change of the grayscale value of the pixel.

[0098] The logic for determining the target region based on the gradient change of pixel gray values ​​is as follows: the gradient of each pixel in the leaf part image is calculated by the Sobel operator, a gradient change threshold is set, pixels with gradients greater than the gradient change threshold are taken as boundary points, and the closed region formed by the boundary points is taken as the target region. The gradient change threshold is set according to expert experience.

[0099] In the gradient image, pixels with gradient values ​​greater than a threshold are marked as boundary points. Binarization can be used to extract these boundary points, and contour detection can be performed using these boundary points to identify all closed regions. The coordinates, area, shape, and other information of the determined target regions are saved as data files, such as CSV or JSON, for subsequent analysis or use.

[0100] The vibration characteristic data specifically refers to the blade vibration amplitude and dominant frequency.

[0101] The logic behind determining the type of blade defect by combining vibration characteristic data of the corresponding blade within a specified time period before shutdown is as follows: The blade vibration amplitude data is extracted from the blade's vibration time history data, and the dominant vibration frequency is analyzed. Based on the degree of deviation of the dominant vibration frequency and the difference in blade vibration amplitude, a structural defect coefficient is calculated to determine whether a structural defect exists on the blade. The specific formula for calculating the structural defect coefficient is as follows:

[0102] ;

[0103] In the formula, Let be the structural defect coefficient of the i-th blade. Let be the dominant vibration frequency of the i-th blade. This is the vibration reference dominant frequency value for the i-th blade;

[0104] It should be noted that the dominant vibration frequency is the core characteristic of blade vibration, and its reference value is... This represents the natural vibration frequency of the blade under normal conditions. When structural defects occur in the blade, such as cracks, surface damage, or internal material deformation, its vibration characteristics will change, leading to a change in the dominant vibration frequency. Deviation from reference frequency value , This indicates the degree of deviation in the main frequency; a larger value means that the blade structure defect is more severe.

[0105] Vibration amplitude is another important characteristic of blade vibration, reflecting the energy and intensity of the vibration. Under normal circumstances, the blade vibration amplitude remains within a relatively stable range. However, if structural defects occur in the blade, such as damage to the blade material or changes in internal stiffness, the vibration amplitude will change significantly. Defined as the degree of abnormality in vibration amplitude, the deviation from the normal amplitude, the larger the value, the more severe the abnormality in blade vibration amplitude;

[0106] Main frequency offset value and amplitude difference These are two complementary indicators: the dominant frequency offset focuses on changes in the blade's vibration mode, while the amplitude difference focuses on abnormal vibration intensity. The combination of the two comprehensively reflects the health status of the blade.

[0107] The logic used to determine whether a blade has structural defects is as follows:

[0108] like At that time, it was determined that the blade had a structural defect;

[0109] like At that time, it was determined that the blade did not have structural defects, and the target area was a stain defect; The threshold for judging structural defects is set based on expert experience.

[0110] The type of structural defect is determined based on the geometric parameters of the target area, including the area, length and circularity of the area, and the type of structural defect includes blade cracks and surface corrosion.

[0111] Since blade cracks are usually long and irregular in shape, with a small area and low roundness, and relatively long crack length, surface corrosion usually presents a certain area and has a smooth shape and high roundness. Therefore, the type of structural defect can be determined by geometric morphology.

[0112] The specific judgment logic is as follows: if the area and roundness of the region are less than the set area threshold and roundness threshold, but the corresponding length is greater than the set length threshold, the structural defect is judged to be a blade crack; if the area and roundness of the region are greater than the set area threshold and roundness threshold, but the corresponding length is less than the set length threshold, the structural defect is judged to be surface corrosion. The area threshold, roundness threshold, and length threshold can be set according to the actual defect situation and expert experience.

[0113] The real-time target adjustment module is used to analyze the degree of structural defects in blades with structural defects, determine the associated detection areas, add the uninspected wind turbines in the associated detection areas to the target set of this round of inspection, and regenerate the inspection path for the remaining wind turbines to be inspected in this round of inspection in order to complete all inspection tasks.

[0114] The logic for determining the specific associated detection area is as follows: Real-time detection and output of blade defect types. If a structural defect is detected in the blade, an associated detection area is defined with the current UAV position as the center. The radius of the associated detection area is determined by the defect severity, specifically by combining the crack length with the area of ​​surface corrosion. The formula for calculating the radius of the associated detection area is as follows:

[0115] ;

[0116] In the formula, The radius of the correlation detection domain, As a reference radius, This represents the total length of the blade's slits. The reference length of the crack. For length influence constant, The constant is the influence of corrosion area. This represents the total corrosion area of ​​the blade. Corrosion reference area;

[0117] It should be noted that the reference radius A baseline value is provided, representing the detection area under normal conditions. This is a basic starting point, through which the size of the associated detection area and the total crack length can be flexibly adjusted. It is directly related to the health of the blade structure; the longer the crack, the more severe the structural damage. The reference crack length is used to compare the difference between the current state and the normal state, by introducing a length influence constant. The change in crack length can be weighted to determine its contribution to the radius of the associated detection area. The setting of this constant can be based on empirical data or experimental results.

[0118] Total corrosion area This reflects the degree of corrosion on the blade surface. The more severe the corrosion, the lower the blade's load-bearing capacity and durability. The corrosion area influence constant is also relevant. This constant is used to adjust the degree of influence of the corrosion area on the radius of the detection field. The impact of corrosion on the health of the blade may vary under different material and environmental conditions; therefore, this constant can be appropriately adjusted based on the specific application scenario and expert experience.

[0119] Where the reference radius The specific setting can be based on the average spacing between the fans, and is generally set to twice the average spacing between the fans.

[0120] Candidate wind turbines in the correlation detection field are added to the inspection target set. Taking the current position of the UAV as the starting point and the landing position as the mission endpoint, the other wind turbines in the inspection target set except for those that have been inspected are taken as path target points. Candidate shortest flight routes are generated based on the nearest neighbor heuristic algorithm, and the candidate shortest flight routes are compared with the current flight range of the UAV.

[0121] If the current flight range of the drone is less than the candidate shortest flight path, the corresponding wind turbine blade inspection will be completed according to the initial inspection path, and the candidate wind turbines in the associated inspection field will be used as the inspection targets for the next round.

[0122] If the drone's current range is not greater than the shortest flight path, then the inspection will be carried out according to the candidate shortest flight path.

[0123] Candidate wind turbines within the associated inspection area are added to the inspection target set to ensure that all potentially defective turbines are included in the inspection scope. This effectively improves the comprehensiveness of the inspection, ensuring that every potential risk point is addressed. The current flight range of the drone is compared with the candidate shortest flight path to ensure that the drone can complete the predetermined inspection targets during the mission, avoiding mission failure due to insufficient flight range. If the flight range is less than the candidate path, the corresponding wind turbine blades are inspected according to the initial inspection path, ensuring safety and controllability within the drone's flight range.

[0124] By setting conditional judgments, the inspection strategy can be dynamically adjusted based on the actual endurance of the drone. When the endurance is insufficient, a safer and controllable method is chosen to continue maintenance. When the endurance is sufficient, an efficient inspection using the candidate shortest flight route is conducted. This flexibility enhances the adaptability and responsiveness of the inspection, reduces unnecessary duplicate inspections by excluding turbines that have already been inspected, optimizes resource utilization, and improves inspection efficiency. Simultaneously, by designating candidate turbines within the associated inspection area as targets for the next round of inspections, a continuous inspection strategy is formed, ensuring smooth transitions in subsequent work.

[0125] Please see Figure 4 The present invention also provides a method for inspecting wind turbine blades based on unmanned aerial vehicles (UAVs). This method is used to control the aforementioned method for inspecting wind turbine blades based on UAVs, and includes:

[0126] Step 1: Collect historical inspection data and working data for a specified period before shutdown for each wind turbine blade in the wind farm. After normalizing the data, construct a health scoring model, score the health status of each blade, and generate a wind turbine health priority queue sorted by health priority.

[0127] Step 2: Based on the initial flight range of the drone and the spacing between adjacent wind turbines in the health priority queue, select wind turbines from the queue to form an inspection target set;

[0128] Step 3: Based on the set of inspection targets, use the nearest neighbor heuristic algorithm to plan the inspection path of the UAV;

[0129] Step 4: Control the drone to inspect the target wind turbines sequentially according to the planned inspection path, and collect image data of the blades during the inspection process. Based on the changes in grayscale values ​​in the images, identify the target area, extract the geometric morphological parameters of the target area, and determine the type of blade defect by combining the vibration characteristic data of the corresponding blades.

[0130] Step 5: Analyze the degree of structural defects in the blades with structural defects, determine the associated detection areas, add the uninspected wind turbines in the associated detection areas to the target set of this round of inspection, and regenerate the inspection path for the remaining wind turbines in this round of inspection to complete all inspection tasks.

[0131] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0132] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0133] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0134] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A wind turbine blade inspection system based on unmanned aerial vehicles (UAVs), characterized in that, Specifically, it includes: The priority analysis module is used to collect historical inspection data of each wind turbine blade in the wind farm and working data for a specified period before shutdown. After normalizing the data, a health scoring model is constructed to score the health status of each blade and generate a wind turbine health priority queue sorted by health priority. The inspection target determination module is used to select wind turbines from the queue to form an inspection target set based on the initial flight range of the UAV and the spacing between adjacent wind turbines in the health priority queue. The inspection route planning module is used to plan the inspection path of the UAV based on the set of inspection targets using the nearest neighbor heuristic algorithm. The blade defect detection module is used to control the drone to inspect the target wind turbine sequentially according to the planned inspection path, and to collect image data of the blades during the inspection process. Based on the changes in grayscale values ​​in the image, the module identifies the target area, extracts the geometric morphology parameters of the target area, and determines the type of blade defect by combining the vibration characteristic data of the corresponding blade. The real-time target adjustment module is used to analyze the degree of structural defects in blades with structural defects, determine the associated detection areas, add the uninspected wind turbines in the associated detection areas to the target set of this round of inspection, and regenerate the inspection path of the remaining wind turbines to be inspected in this round of inspection in order to complete all inspection tasks. The logic for determining the specific associated detection area is as follows: Real-time detection and output of blade defect types. If a structural defect is detected in the blade, an associated detection area is defined with the current UAV position as the center. The radius of the associated detection area is determined by the defect severity, specifically by combining the crack length with the area of ​​surface corrosion. The formula for calculating the radius of the associated detection area is as follows: In the formula, The radius of the correlation detection domain, As a reference radius, This represents the total length of the blade's slits. The reference length of the crack. This is a length-affecting constant. The constant is the influence of corrosion area. This represents the total corrosion area of ​​the blade. Corrosion reference area; Candidate wind turbines in the correlation detection field are added to the inspection target set. Taking the current position of the UAV as the starting point and the landing position as the mission endpoint, the other wind turbines in the inspection target set except for those that have been inspected are taken as path target points. Candidate shortest flight routes are generated based on the nearest neighbor heuristic algorithm, and the candidate shortest flight routes are compared with the current flight range of the UAV. If the current flight range of the drone is less than the candidate shortest flight path, the corresponding wind turbine blade inspection will be completed according to the initial inspection path, and the candidate wind turbines in the associated inspection field will be used as the inspection targets for the next round. If the drone's current range is not less than the shortest flight path, then the inspection will be carried out according to the candidate shortest flight path.

2. The wind turbine blade inspection system based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that: The historical inspection data includes the number of historical faults and the number of years of operation for each blade, and the working data for the specified period before shutdown includes differences in blade vibration amplitude, differences in blade rotation speed, and differences in power generation. Specifically, the differences in vibration amplitude, blade rotation speed, and power generation refer to the absolute differences between the blade vibration amplitude, rotation speed, and power generation within a specified time period before shutdown and the reference vibration amplitude, rotation speed, and power generation at the same wind speed. A health scoring model is constructed to score the health status of each leaf. The formula used to calculate the leaf health status score is as follows: In the formula, Score the health status of the i-th leaf. Let be the normalized value of the historical failure count for the i-th blade. Let be the normalized value of the service life of the i-th blade. Let be the state coefficient of the i-th leaf. and These are the historical scaling constant and the actual scaling constant, respectively, where i is the index of the wind turbine blade; Where the state coefficient of the i-th blade Specifically, it is characterized by the working data of a specified period before shutdown, and the specific formula used for calculation is as follows: In the formula, Let be the normalized value of the vibration amplitude difference of the i-th blade. Let be the normalized value of the rotational speed difference for the i-th blade. Let be the normalized value of the power generation difference of the i-th blade.

3. The wind turbine blade inspection system based on unmanned aerial vehicles (UAVs) according to claim 2, characterized in that: Calculate the average health status score of the blades in each wind turbine, and rank the wind turbines according to the average health status score to form a wind turbine health priority queue. The formula used to calculate the average health status score of each wind turbine is as follows: In the formula, Let be the average health status score corresponding to the j-th wind turbine. For the j-th wind turbine, the health status score of the i-th blade is given. denoted by , where is the number of blades in the wind turbine, and j is the index of the wind turbine within the wind farm; Based on the average health status score, the wind turbines in the wind farm are ranked in descending order to form a wind turbine health priority queue.

4. The wind turbine blade inspection system based on unmanned aerial vehicles (UAVs) according to claim 2, characterized in that: The logic for constructing the inspection target set is as follows: set the take-off and landing positions of the UAV, with the constraint that the length of the preset flight path is not higher than the initial endurance distance, add wind turbines in sequence according to the wind turbine health priority queue to update the preset flight path, and take each wind turbine in the final preset flight path as the inspection wind turbine to form the inspection target set, and take the wind turbines in the wind farm that have not been added to the preset flight path as candidate inspection wind turbines. The preset flight path is the path that starts from the take-off position and ends at the landing position, and passes through each inspection wind turbine in the order of wind turbine addition. The initial endurance distance specifically refers to the maximum flight distance of the UAV when the battery is not consumed.

5. The wind turbine blade inspection system based on unmanned aerial vehicles (UAVs) according to claim 4, characterized in that: The inspection path is the shortest flight route covering each wind turbine in the inspection target set. The specific logic of using the nearest neighbor heuristic algorithm to plan the inspection path of the UAV is as follows: take the take-off position as the mission start point and the landing position as the mission end point, take each wind turbine in the inspection target set as the path target point, and determine the shortest flight route covering each wind turbine in the inspection target set as the inspection path.

6. The wind turbine blade inspection system based on unmanned aerial vehicles (UAVs) according to claim 5, characterized in that: The types of blade defects include structural defects and stain defects; The steps to determine the target area are as follows: perform background separation processing on the collected leaf image data, convert the leaf part in the image into a grayscale image, extract and analyze the grayscale value of each pixel in the grayscale image, and determine the target area based on the gradient change of the grayscale value of the pixel. The logic for determining the target region based on the gradient change of pixel grayscale values ​​is as follows: the gradient of each pixel in the blade image is calculated using the Sobel operator, a gradient change threshold is set, pixels with gradients greater than the gradient change threshold are taken as boundary points, and the closed region formed by the boundary points is taken as the target region; the vibration feature data specifically refers to the blade vibration amplitude and vibration dominant frequency value. The logic behind determining the type of blade defect by combining vibration characteristic data of the corresponding blade within a specified time period before shutdown is as follows: The blade vibration amplitude data is extracted from the blade's vibration time history data, and the dominant vibration frequency is analyzed. Based on the degree of deviation of the dominant vibration frequency and the difference in blade vibration amplitude, a structural defect coefficient is calculated to determine whether a structural defect exists on the blade. The specific formula for calculating the structural defect coefficient is as follows: In the formula, Let be the structural defect coefficient of the i-th blade. Let be the dominant vibration frequency of the i-th blade. Let be the vibration reference dominant frequency value of the i-th blade; The logic used to determine whether a blade has structural defects is as follows: like At that time, it was determined that the blade had a structural defect; like At that time, it was determined that the blade did not have structural defects, and the target area was a stain defect; The threshold for judging structural defects; The type of structural defect is determined based on the geometric parameters of the target area, including the area, length and circularity of the area, and the type of structural defect includes blade cracks and surface corrosion. The specific judgment logic is as follows: if the area and roundness of the region are less than the set area threshold and roundness threshold, but the corresponding length is greater than the set length threshold, the structural defect is judged to be a blade crack; if the area and roundness of the region are greater than the set area threshold and roundness threshold, but the corresponding length is less than the set length threshold, the structural defect is judged to be surface corrosion.

7. A method for inspecting wind turbine blades based on unmanned aerial vehicles (UAVs), characterized in that: The method for inspecting wind turbine blades based on unmanned aerial vehicles (UAVs) is used to control the UAV-based wind turbine blade inspection system according to any one of claims 1-6, comprising: Historical inspection data and working data for a specified period before shutdown of each wind turbine blade in the wind farm are collected. After normalizing the data, a health scoring model is constructed to score the health status of each blade and generate a wind turbine health priority queue sorted by health priority. Based on the initial flight range of the drone and the spacing between adjacent wind turbines in the health priority queue, wind turbines to be inspected are selected from the queue to form an inspection target set; Based on the set of inspection targets, the nearest neighbor heuristic algorithm is used to plan the inspection path of the UAV. The drone is controlled to inspect the target wind turbines sequentially according to the planned inspection path, and image data of the blades is collected during the inspection. Based on the changes in grayscale values ​​in the images, the target area is identified, the geometric morphology parameters of the target area are extracted, and the blade defect type is determined by combining the vibration characteristic data of the corresponding blade. Analyze the degree of structural defects in the blades with defects, determine the associated detection areas, add the uninspected wind turbines in the associated detection areas to the target set of this round of inspection, and regenerate the inspection path for the remaining wind turbines in this round of inspection to complete all inspection tasks.

Images