Power equipment condition monitoring and predictive maintenance management method and system

Abstract

The application is suitable for the field of electric power operation and maintenance, and provides an electric power equipment state monitoring and predictive maintenance management method and system. The method comprises the following steps: collecting initial image data of a fan blade, and establishing a visual reference library, wherein the initial image data is an image of a surface coating of the fan blade in a perfect state; acquiring surface image data of the fan blade in a current state through an image acquisition device carried on a drone; identifying a leading edge part of the blade according to the surface image data to obtain a leading edge image set, and intelligently analyzing and comparing the leading edge image set based on the visual reference library to obtain wear data; and generating a predictive maintenance decision scheme based on the wear data and pose information when the data is collected by the drone. The application improves the accuracy of electric power equipment leading edge state monitoring and the precision of predictive maintenance.

Description

Technical Field

[0001] This invention relates to the field of power operation and maintenance, specifically to a method and system for power equipment condition monitoring and predictive maintenance management. Background Technology

[0002] Power equipment mainly includes two categories: power generation equipment and power supply equipment. Power generation equipment mainly includes power plant boilers, steam turbines, generators, etc., while power supply equipment mainly includes transmission lines of various voltage levels, instrument transformers, etc.

[0003] Monitoring power equipment is fundamental to operation and maintenance management. Taking wind turbines as an example, the blades, as critical power components, directly impact the unit's power generation efficiency and operational safety. The leading edge of the blades is subjected to severe impacts from raindrops, sand particles, and other media during high-speed operation, leading to rain erosion damage on the material surface. Traditional condition monitoring methods rely primarily on periodic manual inspections or high-resolution photographs taken by drones, with professionals assessing for macroscopic defects such as cracks and spalling. This method heavily depends on human experience and struggles to effectively identify and quantify early microscopic rain erosion damage before macroscopic defects have materialized. In many cases, damage is only discovered and addressed after significant performance degradation, resulting in high maintenance costs and a lack of foresight.

[0004] In predictive maintenance management, the current common approach is a preventative maintenance strategy based on fixed cycles. This means that regardless of the actual wear condition of the blades, scheduled downtime inspections or uniform application of protective paint are performed. This "one-size-fits-all" approach has significant drawbacks, easily leading to unnecessary construction work when the coating still provides good protection, resulting in additional downtime losses and material and labor costs. Therefore, this paper proposes a method and system for power equipment condition monitoring and predictive maintenance management, aiming to address the aforementioned problems. Summary of the Invention

[0005] In view of the shortcomings of the existing technology, the purpose of this invention is to provide a method and system for power equipment condition monitoring and predictive maintenance management, so as to solve the problems existing in the above-mentioned background technology.

[0006] This invention is implemented as follows: a method for condition monitoring and predictive maintenance management of power equipment, the method comprising the following steps:

[0007] Initial image data of wind turbine blades are collected, and a visual reference image library is established. The initial image data is an image of the surface coating of the wind turbine blade in an intact state. The surface coating is multi-layered and can change color as the degree of wear increases.

[0008] The surface image data of the wind turbine blades under the current state is obtained by image acquisition equipment mounted on a drone;

[0009] The leading edge of the blade is identified based on surface image data to obtain a leading edge image set, and wear data is obtained by intelligent analysis and comparison of the leading edge image set based on a visual benchmark image library.

[0010] A predictive maintenance decision scheme is generated based on the wear data and the pose information collected by the UAV.

[0011] As a further aspect of the present invention: the step of acquiring surface image data of the wind turbine blades in their current state through an image acquisition device mounted on a drone specifically includes:

[0012] The image acquisition device acquires the current overall image of the wind turbine and identifies and analyzes the state of the wind turbine blades to determine whether the wind turbine blades are stationary. The overall image includes the global field of view of the wind turbine tower and blades.

[0013] When the wind turbine blades are stationary, the spatial pose data of each blade is obtained through overall image recognition. The spatial pose data includes the number of blades, their spatial distribution position, and their attitude angle.

[0014] Based on the spatial pose data and combined with the geometric features of the blade leading edge, a radial scanning path covering all blade leading edge regions is generated.

[0015] Based on the radial scanning path planning, collaborative control commands for the UAV and image acquisition equipment are generated and issued, enabling the UAV to fly along the blade and collect continuous surface image data.

[0016] As a further aspect of the present invention, the method further includes:

[0017] When the wind turbine blades are in motion, acquire wind turbine operating data and environmental data, and establish a distribution model of the wind turbine's wake vortex field;

[0018] The distribution model identifies the wind direction safety zone, and a synchronous scanning path is generated and sent out based on the wind direction safety zone. The synchronous scanning path is an arc-shaped path that is parallel to the blade rotation plane within the wind direction safety zone.

[0019] When the UAV flies along the synchronous scanning path, it acquires the real-time speed information and blade angle information of the wind turbine and establishes a time trajectory model of the wind turbine blades.

[0020] Based on the synchronous scanning path and the time trajectory model, the preset control command of the UAV is obtained by synchronous calculation and issued, so that the UAV can acquire high-speed continuous shooting images of each blade by taking multiple synchronous shots.

[0021] The surface image data is obtained by integrating and processing all the high-speed continuous shooting images.

[0022] As a further aspect of the present invention: the step of obtaining wear data by intelligently analyzing and comparing the leading edge image set based on a visual benchmark image library specifically includes:

[0023] Based on feature point matching, the images in the leading edge image set are registered with the visual reference image library, and the color-changing area of ​​the surface coating is segmented using the edge detection algorithm.

[0024] The color-changing region is converted to the LAB color space and its color channel components are extracted to eliminate the interference of light.

[0025] Based on the preset experimental calibration chromaticity-wear depth mapping relationship, the chromaticity value of each pixel is converted into a specific wear depth value, and a two-dimensional wear depth distribution map is generated.

[0026] Based on the two-dimensional wear depth distribution map, multi-level depth thresholds are set to perform pixel clustering statistics, and quantitative wear area and wear level are output according to the combination rule of area ratio and maximum depth. The wear data includes two-dimensional wear depth distribution map, wear area and wear level.

[0027] As a further aspect of the present invention: the step of generating a predictive maintenance decision scheme based on the wear data and the pose information collected by the UAV specifically includes:

[0028] Based on the pose information of the images collected by the UAV, the wear data is mapped to the corresponding spatial position of the three-dimensional digital model of the wind turbine blade through coordinate transformation;

[0029] Based on the historical and current wear data of each location point on the three-dimensional digital model, the remaining service time of each location point is calculated using a time series prediction algorithm.

[0030] Predictive maintenance decision-making schemes are generated based on the remaining usage time and the wind turbine's power generation plan.

[0031] As a further aspect of the present invention, the method further includes:

[0032] After the maintenance of the wind turbine blades is completed, preview images of the blade surface are pre-captured using image acquisition equipment mounted on a drone;

[0033] The mean, standard deviation, and histogram shape differences of the luminance channels in the LAB color space are compared between the preview image and the image in the visual benchmark library to quantify the lighting conditions of the two images.

[0034] Once the verification is successful, a high-resolution re-inspection image of the repaired surface of the wind turbine blades is captured;

[0035] The high-definition re-inspection images are subjected to quality assessment and integrity verification, and the re-inspection images that pass the verification are synchronously updated to the visual reference image library to establish the latest visual reference for wind turbine blades.

[0036] Another object of the present invention is to provide a power equipment condition monitoring and predictive maintenance management system, the system comprising:

[0037] The benchmark construction module is used to collect initial image data of wind turbine blades and establish a visual benchmark image library. The initial image data is an image of the surface coating of the wind turbine blades in an intact state. The surface coating is multi-layered and can change color as the degree of wear increases.

[0038] The data acquisition module is used to acquire surface image data of the wind turbine blades under their current state through image acquisition equipment mounted on the drone;

[0039] The data analysis module is used to identify the leading edge of the blade based on surface image data to obtain a leading edge image set, and to perform intelligent analysis and comparison of the leading edge image set based on a visual reference image library to obtain wear data.

[0040] The maintenance planning module is used to generate predictive maintenance decision schemes based on the wear data and the pose information collected by the UAV.

[0041] As a further aspect of the present invention: the data acquisition module includes:

[0042] The status recognition unit is used to acquire the current overall image of the wind turbine through the image acquisition device and to identify and analyze the status of the wind turbine blades. It is used to identify whether the wind turbine blades are stationary. The overall image includes the global field of view of the wind turbine tower and blades.

[0043] The pose recognition unit is used to obtain the spatial pose data of each blade through overall image recognition when the wind turbine blade is stationary. The spatial pose data includes the number of blades, spatial distribution position, and pose angle.

[0044] The path generation unit is used to generate a radial scanning path covering all the leading edge parts of the blade based on the spatial pose data and the geometric features of the leading edge of the blade.

[0045] The command issuing unit is used to generate and issue collaborative control commands for the UAV and image acquisition equipment based on the radial scanning path planning, so that the UAV flies along the blade and collects continuous surface image data.

[0046] As a further aspect of the present invention: the data analysis module includes:

[0047] The registration and segmentation unit is used to register the images in the leading edge image set with the visual reference image library based on feature point matching, and to segment the discoloration area of ​​the surface coating using an edge detection algorithm.

[0048] A chromaticity conversion and extraction unit is used to convert the color-changing region to the LAB chromaticity space and extract its chromaticity channel components to eliminate light interference.

[0049] The wear visualization unit is used to convert the chromaticity value of each pixel into a specific wear depth value according to the preset experimental calibration chromaticity-wear depth mapping relationship, and generate a two-dimensional wear depth distribution map;

[0050] The clustering statistics unit is used to perform pixel clustering statistics based on the two-dimensional wear depth distribution map by setting multi-level depth thresholds, and output quantified wear area and wear level according to the combination rule of area proportion and maximum depth. The wear data includes two-dimensional wear depth distribution map, wear area and wear level.

[0051] As a further aspect of the present invention: the system further includes a benchmark update module, which includes:

[0052] The pre-acquisition unit is used to pre-acquire preview images of the blade surface using image acquisition equipment mounted on a drone after the wind turbine blade maintenance is completed;

[0053] The quantization verification unit is used to verify the differences in the mean, standard deviation, and histogram shape of the brightness channel in the LAB color space between the preview image and the image in the visual reference library, and is used to quantify the lighting conditions of the two images.

[0054] The full acquisition unit is used to acquire high-definition re-inspection images of the repaired surface of the wind turbine blades when the verification is passed;

[0055] The evaluation synchronization unit is used to perform quality evaluation and integrity verification on the high-definition re-inspection images, and to synchronously update the re-inspection images that pass the verification to the visual reference image library to establish the latest visual reference for wind turbine blades.

[0056] Compared with the prior art, the beneficial effects of the present invention are:

[0057] This invention pre-establishes a visual reference image library of wind turbine blade surface coatings and utilizes UAV image acquisition equipment to acquire surface image data, achieving accurate identification and image acquisition of the leading edge of the blade. Then, through intelligent analysis and comparison of the leading edge image set based on the visual reference image library, it can effectively identify areas of color change in the coating due to wear at the leading edge and generate accurate wear data. Finally, combined with the pose information acquired by the UAV, a predictive maintenance decision plan for the leading edge is generated, realizing closed-loop management from condition monitoring to maintenance decision-making. In summary, this invention can promptly detect early microscopic damage at the leading edge and achieve quantitative assessment of the degree of leading edge wear through precise analysis of color changes in the leading edge coating. Simultaneously, the use of pose information ensures the accuracy and consistency of image acquisition at the leading edge, greatly improving the reliability of monitoring data. This method overcomes the subjective limitations of traditional manual inspections, avoids the resource waste caused by periodic maintenance, and significantly improves the accuracy of leading edge condition monitoring and the precision of predictive maintenance for power equipment. Attached Figure Description

[0058] Figure 1 This is a flowchart of a method for condition monitoring and predictive maintenance management of power equipment.

[0059] Figure 2 This is a flowchart for acquiring surface image data of wind turbine blades under the current state in a method for monitoring the condition of power equipment and predictive maintenance management.

[0060] Figure 3 This is a flowchart illustrating the intelligent analysis and comparison of leading-edge image sets to obtain wear data in a method for condition monitoring and predictive maintenance management of power equipment.

[0061] Figure 4 This is a flowchart illustrating the generation of predictive maintenance decision schemes in a power equipment condition monitoring and predictive maintenance management method.

[0062] Figure 5 This is a flowchart for updating the visual reference library in a method for monitoring the condition of power equipment and predictive maintenance management.

[0063] Figure 6 This is a schematic diagram of a power equipment condition monitoring and predictive maintenance management system.

[0064] Figure 7 This is a schematic diagram of the data acquisition module in a power equipment condition monitoring and predictive maintenance management system.

[0065] Figure 8 This is a schematic diagram of the data analysis module in a power equipment condition monitoring and predictive maintenance management system.

[0066] Figure 9This is a schematic diagram of the baseline update module in a power equipment condition monitoring and predictive maintenance management system. Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0068] The specific implementation of the present invention will be described in detail below with reference to specific embodiments.

[0069] like Figure 1 As shown in the figure, this invention provides a method for power equipment condition monitoring and predictive maintenance management, the method comprising the following steps:

[0070] S100: Acquire initial image data of the wind turbine blades and establish a visual reference image library. The initial image data is an image of the surface coating of the wind turbine blades in an intact state. The surface coating is multi-layered and can change color as the wear level increases.

[0071] S200 acquires surface image data of the wind turbine blades under their current state through image acquisition equipment mounted on a drone;

[0072] The S300 identifies the leading edge of the blade based on surface image data to obtain a leading edge image set, and performs intelligent analysis and comparison of the leading edge image set based on a visual reference image library to obtain wear data.

[0073] S400, a predictive maintenance decision scheme is generated based on the wear data and the pose information when the UAV collects data.

[0074] It should be noted that the visual benchmark library is a high-definition image database containing wind turbine blades in good condition, acquired from multiple standard angles and under standard lighting conditions. This library provides a wear-free reference benchmark for subsequent image comparison. Each image has a clear spatial location marker to ensure that the comparison is performed at the same location on the blade. The surface coating is a specially formulated intelligent coating with a multi-layered structure. The top layer is a transparent wear-resistant layer, and the layers below it have different, distinct colors. As the wear depth increases, the coating will display a preset color change, thus converting "wear depth" into an intuitive "color signal." The pose information consists of the UAV's spatial position (3D coordinates) and attitude (pitch angle, yaw angle, roll angle) data at the moment the image is acquired. This information is crucial for accurately mapping the 2D image onto the specific location on the 3D model of the blade, ensuring the comparability of images acquired at different times.

[0075] In this embodiment of the invention, a visual reference image library of wind turbine blade surface coatings is pre-established. Surface image data is acquired using a drone image acquisition device, enabling precise identification and image acquisition of the leading edge of the blade. Then, through intelligent analysis and comparison of the leading edge image set based on the visual reference image library, the color change areas of the coating caused by wear at the leading edge can be effectively identified, and accurate wear data can be generated. Finally, combined with the pose information acquired by the drone, a predictive maintenance decision plan for the leading edge is generated, achieving closed-loop management from condition monitoring to maintenance decision-making. In summary, this invention can promptly detect early microscopic damage at the leading edge, and through precise analysis of the color change of the leading edge coating, it achieves a quantitative assessment of the degree of leading edge wear. Simultaneously, the use of pose information ensures the accuracy and consistency of image acquisition at the leading edge, greatly improving the reliability of monitoring data. This method overcomes the subjective limitations of traditional manual inspections, avoids the resource waste caused by periodic maintenance, and significantly improves the accuracy of leading edge condition monitoring and the precision of predictive maintenance for power equipment.

[0076] like Figure 2 As shown, in a preferred embodiment of the present invention, the step of acquiring surface image data of the wind turbine blades in their current state using an image acquisition device mounted on a drone specifically includes:

[0077] S201, The overall image of the wind turbine is acquired through the image acquisition device, and the status of the wind turbine blades is identified and analyzed to identify whether the wind turbine blades are stationary. The overall image includes the global field of view of the wind turbine tower and blades.

[0078] S202, when the wind turbine blades are stationary, the spatial pose data of each blade is obtained through overall image recognition. The spatial pose data includes the number of blades, their spatial distribution position, and their attitude angle.

[0079] S203, Generate a radial scanning path covering all blade leading edge regions based on the spatial pose data and the geometric features of the blade leading edge.

[0080] S204 generates and issues collaborative control commands for the UAV and image acquisition equipment based on the radial scanning path planning, enabling the UAV to fly along the blade and acquire continuous surface image data.

[0081] In practice, there is another method of image acquisition, mainly for acquiring images of wind turbine blades in motion. The specific steps include:

[0082] S212, when the wind turbine blades are in motion, acquire wind turbine operating data and environmental data and establish a distribution model of the wind turbine's wake vortex field;

[0083] S213, Identify the wind direction safety zone according to the distribution model, and generate and send out a synchronous scanning path based on the wind direction safety zone. The synchronous scanning path is an arc-shaped path that is parallel to the blade rotation plane within the wind direction safety zone.

[0084] S214: When the UAV flies according to the synchronous scanning path, it acquires the real-time speed information and blade angle information of the wind turbine and establishes a time trajectory model of the wind turbine blades.

[0085] S215, based on the synchronous scanning path and the time trajectory model, the preset control command of the UAV is obtained by synchronous calculation and issued, so that the UAV can acquire high-speed continuous shooting images of each blade by taking multiple synchronous shots, and then integrate and process all the high-speed continuous shooting images to obtain the surface image data.

[0086] In this embodiment of the invention, the intelligent closed-loop process of image acquisition when the wind turbine blades are stationary is first described. First, the drone does not fly directly close to the blades, but instead captures a global image of the wind turbine from a safe distance using a wide-angle lens. The system's visual recognition algorithm locates the tower and blades in this image and determines whether the wind turbine is completely stationary or in motion by analyzing the movement of blade edge pixels in multiple consecutive frames. Once stationary is confirmed, further refined image recognition is performed on each blade. For example, the blade number can be determined using specific marker points, and a multi-view geometric algorithm is used to calculate the specific orientation, pitch angle, and torsion of each blade in three-dimensional space. Based on this high-precision spatial pose data, the path planning algorithm no longer generates a simple straight path, but instead designs a "radial scanning path" for each blade. This path begins at the connection between the blade root and the hub and extends strictly along the curvature of the blade's leading edge towards the blade tip, ensuring that the flight trajectory maintains a stable optimal imaging distance from the blade's leading edge. Finally, this path was translated into a series of flight control commands, which not only controlled the UAV's three-dimensional spatial position but also synchronously controlled the gimbal's pitch and yaw axes through underlying protocols. This ensured that the camera lens remained perpendicular to the local tangent on the blade surface, allowing for the acquisition of continuous, uniform, and clear images of the leading edge surface during the UAV's uniform flight. The entire process required no manual intervention, significantly improving the efficiency and consistency of data acquisition.

[0087] In addition, when the system detects blade rotation, it first acquires real-time wind turbine operating data (including speed and blade pitch angle) and environmental data (including wind direction and wind speed) from the wind turbine's SCADA system via a wireless data link (such as 4G / 5G). This data is then input into a pre-set aerodynamic wake model (such as the Jensen model) to calculate in real-time the areas of maximum vortex intensity and danger behind the wind turbine, as well as relatively stable safe zones. Within this safe zone, the system plans a "synchronous scanning path," which is not a straight line but an arc parallel to the blade tip's rotation trajectory, allowing the UAV to maintain relative lateral movement with the blade tip along this path. It is important to note that the UAV flight control system needs to perform high-frequency clock synchronization and data interaction with the wind turbine's main controller to receive precise blade angle information in real time. Based on this, the system can predict the spatial location point the blade tip will pass through within the next few hundred milliseconds (obtained through a time trajectory model). The flight control system pre-calculates commands, driving the UAV to enter from a safe area at maximum speed at the appropriate time, thus achieving speed synchronization with the target blade tip at a fleeting moment. Within this extremely short synchronization window, the camera is controlled to capture high-speed continuous shots. The UAV then quickly desynchronizes, returns to a safe position, and waits for the next opportunity to repeat the process on another blade. Finally, the high-speed image sequence captured multiple times simultaneously is integrated into a complete, seamless surface image using image registration and stitching algorithms. This method cleverly transforms the disadvantages of UAV dynamic performance into the advantage of precise spatiotemporal prediction, making it a safe and effective method for achieving high-quality imaging of rotating blades.

[0088] like Figure 3 As shown in the preferred embodiment of the present invention, the step of obtaining wear data by intelligent analysis and comparison of the leading edge image set based on the visual reference image library specifically includes:

[0089] S301, based on feature point matching, registers the images in the leading edge image set with the visual reference image library, and uses the edge detection algorithm to segment the color-changing area of ​​the surface coating;

[0090] S302, the color-changing region is converted to the LAB color space and its color channel components are extracted to eliminate the interference of light.

[0091] S303 converts the chromaticity value of each pixel into a specific wear depth value according to the preset experimental calibration chromaticity-wear depth mapping relationship, and generates a two-dimensional wear depth distribution map;

[0092] S304, based on the two-dimensional wear depth distribution map, set multi-level depth thresholds to perform pixel clustering statistics, and output quantified wear area and wear level according to the combination rule of area ratio and maximum depth. The wear data includes two-dimensional wear depth distribution map, wear area and wear level.

[0093] This invention describes an intelligent analysis process for accurately quantifying wear levels from acquired images, primarily by converting visual color changes into precise physical wear data. First, based on feature point registration techniques (such as the ORB algorithm), the leading-edge image acquired on-site is pixel-level aligned with a standard image at the same position and angle in a visual reference library, thus eliminating errors caused by shooting angle deviations. Then, an edge detection algorithm is used to identify the contours of the coating area, and image differencing techniques are used to segment out the "color-changing areas" with discrepancies. Furthermore, to eliminate interference from external light on color judgment, the pixels in the color-changing areas are converted from the RGB color space to the LAB color space, and the a (red-green) and b (blue-yellow) chromaticity channel components are extracted separately for analysis. This utilizes the characteristic that chromaticity information in the LAB space is independent of brightness, thus accurately reflecting the true color of the coating. Then, the system uses a pre-established "chromaticity-wear depth" mapping database to convert the a and b values ​​of each pixel into a specific wear depth estimate, thereby obtaining a two-dimensional distribution map where grayscale values ​​represent wear depth. The system sets multiple depth thresholds on this distribution map, counts the number of pixels exceeding each threshold, calculates the wear area of ​​each level and its proportion of the total area; then, by combining the maximum wear depth value and the area proportion of each level, a set of preset combination rules is used to obtain quantified wear area data and a comprehensive wear level evaluation. The above method can provide accurate and reliable decision-making basis for subsequent predictive maintenance.

[0094] like Figure 4 As shown, in a preferred embodiment of the present invention, the step of generating a predictive maintenance decision scheme based on the wear data and the pose information collected by the UAV specifically includes:

[0095] S401, based on the pose information of the images collected by the UAV, maps the wear data to the corresponding spatial position of the three-dimensional digital model of the wind turbine blade through coordinate transformation;

[0096] S402, based on the historical and current wear data of each location point on the three-dimensional digital model, uses a time series prediction algorithm to calculate the remaining service time of each location point;

[0097] S403 generates predictive maintenance decision schemes based on remaining usage time and wind turbine power generation plans.

[0098] In this embodiment of the invention, firstly, the high-precision pose information recorded by the UAV during image acquisition is utilized. A coordinate transformation matrix is ​​used to accurately map the wear points identified in the two-dimensional image onto the corresponding surface of the three-dimensional digital model of the wind turbine blade, establishing a three-dimensional wear map containing spatial distribution. Subsequently, historical wear data for that location (such as wear depth recorded in previous inspection cycles) is retrieved, and a time series prediction algorithm (such as an LSTM long short-term memory network) is used to fit and predict the wear trend. The number of cycles or time required for that point to reach the critical failure thickness at the current wear rate is calculated, i.e., the remaining service life. For example, if the current wear depth at a certain location is 200 μm, and historical data shows it deepens by 10 μm per month, assuming the critical failure thickness is 500 μm, the remaining service life of that point can be predicted to be 30 months. Finally, the maintenance decision engine integrates the remaining service life prediction results of all locations and performs multi-dimensional optimization matching with the wind farm's power generation plan (such as wind speed predictions for future months and grid dispatch requirements). The optimal predictive maintenance decision scheme is generated through the calculation of constraints and objective functions.

[0099] like Figure 5 As shown, in a preferred embodiment of the present invention, the power equipment condition monitoring and predictive maintenance management method further includes:

[0100] S501, after the maintenance of the wind turbine blades is completed, uses an image acquisition device mounted on a drone to pre-acquire preview images of the blade surface;

[0101] S502, verify the difference in mean, standard deviation and histogram shape of the brightness channel in the LAB color space between the preview image and the image in the visual reference library, in order to quantify the lighting conditions of the two images;

[0102] S503: When the verification is passed, a high-definition re-inspection image of the repaired surface of the wind turbine blades is captured;

[0103] S504, perform quality assessment and integrity verification on the high-definition re-inspection image, and synchronously update the re-inspection image that passes the verification to the visual reference image library to establish the latest visual reference for wind turbine blades.

[0104] In this embodiment of the invention, after maintenance is completed, the drone first flies to a preset location to collect a low-resolution preview image. The system then converts this preview image and the corresponding reference image in the visual reference library to the LAB color space, extracts the L (luminance) channel for statistical analysis, and calculates the mean difference (ΔL), standard deviation ratio (σ1 / σ2), and histogram distribution similarity (e.g., using the Barthel coefficient) of the L channels of the two images to accurately quantify the difference in ambient lighting conditions between the two shots. If all lighting difference indicators are below a preset threshold, the lighting conditions are considered consistent and the comparison is valid. Only then will the drone be triggered to collect high-definition re-inspection images according to standard operating procedures. After acquiring the high-definition re-inspection images, an automated quality assessment is performed, mainly including checking image clarity, whether the target area is fully covered, and whether there is significant distortion, to ensure image usability. After passing the above verification, the system automatically adds these images to the visual reference library to replace the old reference images, thereby completing the recording of the latest health status of the leaves. The above method completely eliminates the reference distortion caused by coating updates, realizes the dynamic evolution of visual reference, and enables the predictive maintenance system to have self-updating capability.

[0105] like Figure 6 As shown in the figure, this embodiment of the invention also provides a power equipment condition monitoring and predictive maintenance management system, the system comprising:

[0106] The benchmark construction module 100 is used to collect initial image data of wind turbine blades and establish a visual benchmark image library. The initial image data is an image of the surface coating of the wind turbine blades in an intact state. The surface coating is multi-layered and can change color as the degree of wear increases.

[0107] The data acquisition module 200 is used to acquire surface image data of the wind turbine blades under the current state through the image acquisition equipment mounted on the UAV;

[0108] The data analysis module 300 is used to identify the leading edge of the blade based on surface image data to obtain a leading edge image set, and to perform intelligent analysis and comparison of the leading edge image set based on a visual reference image library to obtain wear data.

[0109] The maintenance planning module 400 is used to generate predictive maintenance decision schemes based on the wear data and the pose information collected by the UAV.

[0110] In this embodiment of the invention, a complete predictive maintenance closed loop for power equipment is constructed through four core modules. The benchmark construction module 100 establishes a visual benchmark image library of the intelligent coating in the intact state of the blades, providing a standard reference for subsequent comparisons. The data acquisition module 200 dynamically acquires blade surface image data using an imaging device mounted on a UAV. The data analysis module 300 intelligently processes the acquired leading-edge images and generates quantitative wear data through comparison and analysis with the benchmark image library. The maintenance planning module 400 ultimately integrates the wear data and spatial pose information to generate a decision-making scheme that includes maintenance timing, location, and resources, achieving intelligent management from condition monitoring to maintenance execution.

[0111] like Figure 7 As shown, in a preferred embodiment of the present invention, the data acquisition module 200 includes:

[0112] The status recognition unit 201 is used to acquire the current overall image of the wind turbine through the image acquisition device and to identify and analyze the status of the wind turbine blades. It is used to identify whether the wind turbine blades are stationary. The overall image includes the global field of view of the wind turbine tower and blades.

[0113] The pose recognition unit 202 is used to obtain the spatial pose data of each blade through overall image recognition when the wind turbine blade is stationary. The spatial pose data includes the number of blades, spatial distribution position and pose angle.

[0114] The path generation unit 203 is used to generate a radial scanning path covering all the leading edge parts of the blade based on the spatial pose data and the geometric features of the leading edge of the blade.

[0115] The command issuing unit 204 is used to generate and issue collaborative control commands for the UAV and image acquisition equipment based on the radial scanning path planning, so that the UAV flies along the blade and collects continuous surface image data.

[0116] like Figure 8 As shown, in a preferred embodiment of the present invention, the data analysis module 300 includes:

[0117] The registration and segmentation unit 301 is used to register the images in the leading edge image set with the visual reference image library based on feature point matching, and to segment the discoloration area of ​​the surface coating using an edge detection algorithm.

[0118] The chromaticity conversion and extraction unit 302 is used to convert the color-changing region to the LAB chromaticity space and extract its chromaticity channel components to eliminate the interference of light.

[0119] Wear visualization unit 303 is used to convert the chromaticity value of each pixel into a specific wear depth value according to the preset experimental calibration chromaticity-wear depth mapping relationship, and generate a two-dimensional wear depth distribution map;

[0120] The clustering statistics unit 304 is used to perform pixel clustering statistics based on the two-dimensional wear depth distribution map by setting multi-level depth thresholds, and output quantified wear area and wear level according to the combination rule of area proportion and maximum depth. The wear data includes two-dimensional wear depth distribution map, wear area and wear level.

[0121] like Figure 9 As shown, in a preferred embodiment of the present invention, the power equipment condition monitoring and predictive maintenance management system further includes a baseline update module 500, which includes:

[0122] The pre-acquisition unit 501 is used to pre-acquire preview images of the blade surface using an image acquisition device mounted on a drone after the wind turbine blade maintenance is completed;

[0123] The quantization verification unit 502 is used to verify the differences in the mean, standard deviation and histogram shape of the brightness channel of the preview image and the image in the visual reference library in the LAB color space, and is used to quantify the lighting conditions of the two images.

[0124] The full acquisition unit 503 is used to acquire high-definition re-inspection images of the repaired surface of the wind turbine blades when the verification is passed;

[0125] The evaluation synchronization unit 504 is used to perform quality evaluation and integrity verification on the high-definition re-inspection image, and to synchronously update the re-inspection image that has passed the verification to the visual reference image library to establish the latest visual reference for the wind turbine blade.

[0126] The above description only details the preferred embodiments of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0127] It should be understood that although the steps in the flowcharts of the various embodiments of the present invention are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the various embodiments may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least a portion of the sub-steps or stages of other steps.

[0128] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0129] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the disclosure in the specification and embodiments. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the claims.

Claims

1. A method for condition monitoring and predictive maintenance management of power equipment, characterized in that, The method includes the following steps: Initial image data of wind turbine blades are collected, and a visual reference image library is established. The initial image data is an image of the surface coating of the wind turbine blade in an intact state. The surface coating is multi-layered and can change color as the degree of wear increases. The surface image data of the wind turbine blades under the current state is obtained by image acquisition equipment mounted on a drone; The leading edge of the blade is identified based on surface image data to obtain a leading edge image set, and wear data is obtained by intelligent analysis and comparison of the leading edge image set based on a visual benchmark image library. A predictive maintenance decision scheme is generated based on the wear data and the pose information collected by the UAV. The step of obtaining wear data by intelligently analyzing and comparing the leading edge image set based on the visual benchmark image library specifically includes: Based on feature point matching, the images in the leading edge image set are registered with the visual reference image library, and the color-changing area of ​​the surface coating is segmented using the edge detection algorithm. The color-changing region is converted to the LAB color space and its color channel components are extracted to eliminate the interference of light. Based on the preset experimental calibration chromaticity-wear depth mapping relationship, the chromaticity value of each pixel is converted into a specific wear depth value, and a two-dimensional wear depth distribution map is generated. Based on the two-dimensional wear depth distribution map, multi-level depth thresholds are set to perform pixel clustering statistics, and quantitative wear area and wear level are output according to the combination rule of area ratio and maximum depth. The wear data includes two-dimensional wear depth distribution map, wear area and wear level.

2. The method for condition monitoring and predictive maintenance management of power equipment according to claim 1, characterized in that, The step of acquiring surface image data of the wind turbine blades under their current state using an image acquisition device mounted on a drone specifically includes: The image acquisition device acquires the current overall image of the wind turbine and identifies and analyzes the state of the wind turbine blades to determine whether the wind turbine blades are stationary. The overall image includes the global field of view of the wind turbine tower and blades. When the wind turbine blades are stationary, the spatial pose data of each blade is obtained through overall image recognition. The spatial pose data includes the number of blades, their spatial distribution position, and their attitude angle. Based on the spatial pose data and combined with the geometric features of the blade leading edge, a radial scanning path covering all blade leading edge regions is generated. Based on the radial scanning path planning, collaborative control commands for the UAV and image acquisition equipment are generated and issued, enabling the UAV to fly along the blade and collect continuous surface image data.

3. The method for condition monitoring and predictive maintenance management of power equipment according to claim 2, characterized in that, The method further includes: When the wind turbine blades are in motion, acquire wind turbine operating data and environmental data, and establish a distribution model of the wind turbine's wake vortex field; The distribution model identifies the wind direction safety zone, and a synchronous scanning path is generated and sent out based on the wind direction safety zone. The synchronous scanning path is an arc-shaped path that is parallel to the blade rotation plane within the wind direction safety zone. When the UAV flies along the synchronous scanning path, it acquires the real-time speed information and blade angle information of the wind turbine and establishes a time trajectory model of the wind turbine blades. Based on the synchronous scanning path and the time trajectory model, the preset control command of the UAV is obtained by synchronous calculation and issued, so that the UAV can acquire high-speed continuous shooting images of each blade by taking multiple synchronous shots. The surface image data is obtained by integrating and processing all the high-speed continuous shooting images.

4. The method for condition monitoring and predictive maintenance management of power equipment according to claim 1, characterized in that, The step of generating a predictive maintenance decision scheme based on the wear data and the pose information collected by the UAV specifically includes: Based on the pose information of the images collected by the UAV, the wear data is mapped to the corresponding spatial position of the three-dimensional digital model of the wind turbine blade through coordinate transformation; Based on the historical and current wear data of each location point on the three-dimensional digital model, the remaining service time of each location point is calculated using a time series prediction algorithm. Predictive maintenance decision-making schemes are generated based on the remaining usage time and the wind turbine's power generation plan.

5. The method for condition monitoring and predictive maintenance management of power equipment according to claim 1, characterized in that, The method further includes: After the maintenance of the wind turbine blades is completed, preview images of the blade surface are pre-captured using image acquisition equipment mounted on a drone; The mean, standard deviation, and histogram shape differences of the luminance channels in the LAB color space are compared between the preview image and the image in the visual benchmark library to quantify the lighting conditions of the two images. Once the verification is successful, a high-resolution re-inspection image of the repaired surface of the wind turbine blades is captured; The high-definition re-inspection images are subjected to quality assessment and integrity verification, and the re-inspection images that pass the verification are synchronously updated to the visual reference image library to establish the latest visual reference for wind turbine blades.

6. A power equipment condition monitoring and predictive maintenance management system, characterized in that, The system includes: The benchmark construction module is used to collect initial image data of wind turbine blades and establish a visual benchmark image library. The initial image data is an image of the surface coating of the wind turbine blades in an intact state. The surface coating is multi-layered and can change color as the degree of wear increases. The data acquisition module is used to acquire surface image data of the wind turbine blades under their current state through image acquisition equipment mounted on the drone; The data analysis module is used to identify the leading edge of the blade based on surface image data to obtain a leading edge image set, and to perform intelligent analysis and comparison of the leading edge image set based on a visual reference image library to obtain wear data. The maintenance planning module is used to generate predictive maintenance decision schemes based on the wear data and the pose information collected by the UAV. The data analysis module includes: The registration and segmentation unit is used to register the images in the leading edge image set with the visual reference image library based on feature point matching, and to segment the discoloration area of ​​the surface coating using an edge detection algorithm. A chromaticity conversion and extraction unit is used to convert the color-changing region to the LAB chromaticity space and extract its chromaticity channel components to eliminate light interference. The wear visualization unit is used to convert the chromaticity value of each pixel into a specific wear depth value according to the preset experimental calibration chromaticity-wear depth mapping relationship, and generate a two-dimensional wear depth distribution map; The clustering statistics unit is used to perform pixel clustering statistics based on the two-dimensional wear depth distribution map by setting multi-level depth thresholds, and output quantified wear area and wear level according to the combination rule of area proportion and maximum depth. The wear data includes two-dimensional wear depth distribution map, wear area and wear level.

7. The power equipment condition monitoring and predictive maintenance management system according to claim 6, characterized in that, The data acquisition module includes: The status recognition unit is used to acquire the current overall image of the wind turbine through the image acquisition device and to identify and analyze the status of the wind turbine blades. It is used to identify whether the wind turbine blades are stationary. The overall image includes the global field of view of the wind turbine tower and blades. The pose recognition unit is used to obtain the spatial pose data of each blade through overall image recognition when the wind turbine blade is stationary. The spatial pose data includes the number of blades, spatial distribution position, and pose angle. The path generation unit is used to generate a radial scanning path covering all the leading edge parts of the blade based on the spatial pose data and the geometric features of the leading edge of the blade. The command issuing unit is used to generate and issue collaborative control commands for the UAV and image acquisition equipment based on the radial scanning path planning, so that the UAV flies along the blade and collects continuous surface image data.

8. The power equipment condition monitoring and predictive maintenance management system according to claim 6, characterized in that, The system also includes a baseline update module, which includes: The pre-acquisition unit is used to pre-acquire preview images of the blade surface using image acquisition equipment mounted on a drone after the wind turbine blade maintenance is completed; The quantization verification unit is used to verify the differences in the mean, standard deviation, and histogram shape of the brightness channel in the LAB color space between the preview image and the image in the visual reference library, and is used to quantify the lighting conditions of the two images. The full acquisition unit is used to acquire high-definition re-inspection images of the repaired surface of the wind turbine blades when the verification is passed; The evaluation synchronization unit is used to perform quality evaluation and integrity verification on the high-definition re-inspection images, and to synchronously update the re-inspection images that pass the verification to the visual reference image library to establish the latest visual reference for wind turbine blades.

Images