A method for controlling the pre-embedded airfoil end face wrinkle of a wind power blade root

Abstract

This invention relates to the field of wind power generation technology. It discloses a method for controlling wrinkles on the pre-embedded blade end face at the root of a wind turbine blade. The method includes: calibrating material properties to obtain calibration parameters; acquiring macroscopic point cloud data and aligning it with a DEM file to form registration data; extracting features based on the calibration parameters to obtain a first feature set; and fusing the registration data with the first feature set to obtain a first dataset. This invention acquires material properties and surface morphology through a combination of nanoindentation and laser scanning, and dynamically adjusts the point cloud denoising threshold using a stiffness coefficient. This preserves the micron-level morphological details of high-stiffness materials while suppressing noise in low-stiffness materials, improving data quality. Finally, it enhances feature representation through fractal kernel convolution and Riemannian manifold alignment, segments defect regions through three-dimensional fractal dimension differences, and achieves defect visualization using a transfer function.

Description

Technical Field

[0001] This invention relates to the field of wind power generation technology, and more specifically, to a method for controlling the wrinkles on the end face of the blade pre-embedded at the root of a wind turbine blade. Background Technology

[0002] Wrinkles on the embedded blade end face at the root of wind turbine blades refer to wrinkles that appear on the end face of the embedded blade profile during the manufacturing process of the critical connection between the blade and the hub. These wrinkles can be caused by uneven fiberglass fabric laying, improper curing processes, mold defects, or uneven core material transitions. This defect reduces blade strength and fatigue life, and may lead to surface tangential cracks, internal structural delamination, or even fracture. The blade root must withstand cyclic alternating stress; wrinkles exacerbate stress concentration, gradually expanding during operation and causing blade failure. Typical causes include: uneven fiberglass laying, uneven mold heating, thermal stress from resin curing, and improper design of the core material thickness transition zone.

[0003] Existing control methods have significant shortcomings. First, traditional point cloud denoising techniques often use fixed thresholds, which are difficult to adapt to the stiffness characteristics of different materials. This results in the inability to accurately identify and eliminate noise under complex working conditions, thus affecting the accuracy of the structural model. Second, there are deficiencies in multimodal data fusion. Existing technologies typically process acoustic, optical, and other data in isolation, lacking an effective spatial registration mechanism, which causes inconsistencies between data and limits the comprehensive analysis of wrinkle features. In addition, feature extraction methods rely excessively on a single signal source or simple parameters, failing to fully integrate multidimensional information. These shortcomings limit the adaptability, accuracy, and comprehensiveness of the control methods, making it difficult to achieve effective control of wrinkles on the blade root end face. Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present invention provide a method for controlling the wrinkles on the end face of the blade pre-embedded at the root of a wind turbine blade.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A method for controlling the wrinkles on the end face of the blade pre-embedded at the root of a wind turbine blade, the method comprising: S101: Preprocess the pre-embedded blade end face area at the root of the blade to obtain material properties, scan the preprocessed end face area and calculate the surface roughness of the end face area to obtain a DEM file; S102: The material properties are calibrated to obtain calibration parameters. Macroscopic point cloud data is acquired and aligned with the DEM file to form registration data. Feature extraction is performed based on the calibration parameters to obtain the first feature set. The registration data is fused with the first feature set to obtain the first dataset. S103: Extract the abnormal regions from the first dataset to obtain the first matrix. Calculate the defect parameters using the first matrix, classify and quantify the defect parameters, and obtain the defect score. S104: Construct feature vectors based on the first dataset and defect parameters, enhance the feature vectors to form enhanced feature vectors, and identify defect areas.

[0006] Further, the scanning preprocessed end face region and the calculation of the surface roughness of the end face region include: The scanning end face area generates micro-point cloud data, which is then optimized to obtain optimized point cloud data. The surface roughness of the cross-sectional area is then calculated based on the optimized point cloud data.

[0007] Furthermore, the optimization of micro-viewpoint cloud data includes: The stiffness coefficient is calculated based on the material properties. The removal threshold is adjusted using the stiffness coefficient. The average distance of the neighboring points is calculated for each point. The average distance is compared with the adjusted removal threshold to obtain the denoised point cloud data. Calculate the local gradient of the denoised point cloud data, determine the interpolation weight coefficients based on the local gradient, and perform adaptive interpolation to obtain optimized point cloud data.

[0008] Furthermore, the feature extraction based on calibration parameters includes: Acoustic signals are acquired, preprocessed, and feature maps are obtained by combining calibration parameters. Based on the feature maps, three types of feature parameters are calculated to obtain the first feature set. Among them, the three types of characteristic parameters are: main frequency energy concentration. Time-domain waveform kurtosis and frequency domain entropy The calculation formulas are as follows: , , , In the formula: This represents the elastic modulus weighting coefficient, where E represents the measured elastic modulus. The center frequency of the k-th frequency component is... Represents frequency Spectral energy density at that location The total number of frequency components in the spectrum. This represents the set of defect-sensitive frequency bands. Represents the entire frequency band set. For frequency band indexing, This is the lowest frequency of frequency band b. This is the highest frequency of frequency band b. The magnitude coefficient of the exponential term. The decay rate of the exponential term. The amplitude coefficient of the sine term. Here, t is the frequency coefficient of the sinusoidal term, t is the time variable, and e is the natural constant; A second matrix is ​​constructed based on the concentration of the main frequency energy and the kurtosis of the time-domain waveform.

[0009] Furthermore, the fusion of the registration data with the first feature set includes: Align the registration data with the first feature set, and then map the first feature set onto the coordinate system of the registration data to obtain the first dataset and calculate the registration error.

[0010] Furthermore, the classification and quantification of defect parameters includes: The physical characteristics of the defect parameters are verified, and the defects are classified and the severity score is calculated based on their geometric shape and the threshold of the characteristic parameters. A quantitative table is generated by integrating defect type, size, coordinates, and severity score.

[0011] Furthermore, the enhancement processing of the feature vector includes: The feature vectors are reconstructed, and the reconstructed feature vectors are aligned with each other to obtain aligned multi-dimensional features. Generate a fractal kernel, and perform a convolution operation between the fractal kernel and the aligned multi-dimensional features to form enhanced features; The enhanced feature vector is obtained by concatenating the enhanced feature and the feature vector and performing a Hadamard product operation.

[0012] Furthermore, the specific steps for identifying the defective region are as follows: The enhanced feature vectors are mapped into three-dimensional space to obtain a three-dimensional feature point cloud. The fractal dimension of the neighborhood of each point is calculated. Based on the difference in fractal dimension, the defect region is segmented to obtain a defect region mask. A 3D mesh is constructed, the gradient field and normal vector of each mesh are calculated, and each mesh is marked by a defect region mask to obtain a 3D defect map. A transfer function is designed and used to visualize the 3D defect map to obtain the defect region.

[0013] An electronic device includes a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the control method for pre-embedded blade end face wrinkles at the root of the wind turbine blade as described above.

[0014] A computer-readable storage medium storing a computer program, which, when executed, implements the control method for pre-embedded blade end face wrinkles at the root of any of the above-described wind turbine blades.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This application discloses a method for controlling the wrinkles on the end face of the pre-embedded blade at the root of a wind turbine blade, comprising: calibrating the material properties to obtain calibration parameters, acquiring macroscopic point cloud data and aligning it with a DEM file to form registration data, extracting features based on the calibration parameters to obtain a first feature set, and fusing the registration data with the first feature set to obtain a first dataset. This invention acquires material properties and surface morphology through the collaborative acquisition of nanoindentation and laser scanning, and dynamically adjusts the point cloud denoising threshold using stiffness coefficients. This preserves the micron-level morphological details of high-stiffness materials while suppressing noise in low-stiffness materials, thus improving data quality. Secondly, acoustic signals and point cloud data are registered and fused to construct a multi-dimensional feature set, which is then used in conjunction with a region growing algorithm to extract abnormal regions. Furthermore, a defect parameter classification and quantification scoring mechanism, combined with geometric morphology and physical features, forms a standardized defect assessment system to identify defect types such as wrinkles and holes. Finally, feature expression is enhanced through fractal kernel convolution and Riemannian manifold alignment, and defect regions are segmented using three-dimensional fractal dimension differences. Defect visualization is achieved using a transfer function, forming a closed-loop process from data acquisition to defect identification, significantly improving detection efficiency and reliability. Attached Figure Description

[0016] Figure 1 A flowchart of a method for controlling the end face wrinkles of a pre-embedded blade at the root of a wind turbine blade, provided by the present invention; Figure 2 A schematic diagram of the structure of an electronic device provided by the present invention; Figure 3 This is a schematic diagram of the structure of a computer-readable storage medium provided by the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Example 1:

[0019] Please see Figure 1 As shown in the figure, this embodiment discloses a method for controlling the wrinkles on the end face of the blade profile pre-embedded at the root of a wind turbine blade, the method comprising: S101: Preprocess the pre-embedded blade end face area at the root of the blade to obtain material properties, scan the preprocessed end face area and calculate the surface roughness of the end face area to obtain a DEM file; In this embodiment, the pretreatment of the pre-embedded leaf-shaped end face area at the leaf root includes: Using an MTS NanoIndenter G200 nanoindenter, test points (25 points in total) were divided into 5mm×5mm grids on the end face area. A constant load of 0.5mN was applied to each point, held for 10 seconds, and then unloaded to obtain the elastic modulus and hardness as material properties.

[0020] As one specific implementation, the step of scanning the preprocessed end face region and calculating the surface roughness of the end face region includes: The scanning end face area generates micro-point cloud data, which is then optimized to obtain optimized point cloud data. The surface roughness of the cross-sectional area is then calculated based on the optimized point cloud data.

[0021] The optimization of micro-viewpoint cloud data includes: The stiffness coefficient k is calculated based on the material properties, i.e., k=E / ρ, where ρ is the material density. The stiffness coefficient is used to adjust the rejection threshold. The average distance of the neighboring points is calculated for each point. The average distance is compared with the adjusted rejection threshold to obtain the denoised point cloud data. The specific formula for adjusting the rejection threshold using the stiffness coefficient is as follows: , In the formula: =0.03 μm As the baseline threshold, This is the stiffness adjustment coefficient, with a value of 0.01. μm -1 The aim is to use a more lenient rejection threshold when dealing with high-stiffness materials to preserve micron-level morphological details, and a more stringent threshold when dealing with low-stiffness materials to suppress noise.

[0022] If the average distance is -0.1μm > If the value is 0, it is considered an outlier and is removed.

[0023] Calculate the local gradient of the denoised point cloud data , This represents the partial derivative in the x-direction. This represents the partial derivative in the y-direction. The interpolation weight coefficients are determined based on the local gradient, and adaptive interpolation is performed to obtain optimized point cloud data.

[0024] It should be noted that the formula for calculating the interpolation weighting coefficients is as follows: In the formula: This is the gradient decay coefficient, with a value of 0.01μm-1. Points with larger gradients are assigned lower weights to suppress over-smoothing, while points with smaller gradients are assigned higher weights to improve interpolation smoothness.

[0025] For the missing regions, a weighted average interpolation formula is used: , In the formula: Indicates the elevation of neighboring points; Meanwhile, the calculation steps for the surface roughness of the cross-sectional area are as follows: Quantifying surface roughness using the fractal dimension method: , , In the formula: For conversion factors, This represents the fractal dimension of an ideal smooth surface, with a value of 2. This is the scale parameter.

[0026] A DEM is a model that stores ground (or object surface) elevation information in digital form, representing the elevation value of each point in space through regular or irregular grids. In blade inspection, it specifically refers to the three-dimensional surface model of the blade root end face, containing X (length), Y (width), and Z (height / depth) coordinates.

[0027] S102: The material properties are calibrated to obtain calibration parameters. Macroscopic point cloud data is acquired and aligned with the DEM file to form registration data. Feature extraction is performed based on the calibration parameters to obtain the first feature set. The registration data is fused with the first feature set to obtain the first dataset. It should be noted that the calibration of material properties specifically involves: configuring the lidar resolution to 0.05mm and the acoustic sensor frequency to 5MHz to ensure matching with the material properties; using standard test blocks for online calibration; adjusting the laser power to ensure that the signal-to-noise ratio of point cloud acquisition is ≥30dB; and calibrating the acoustic wave transmission-reception delay time error to ≤0.1μs.

[0028] Standard test blocks are standardized specimens with known defect parameters (such as crack length, pore volume, inclusion location, etc.). Their materials, geometry, defect types, and dimensions are all precision-machined and calibrated by third-party metrology institutions, conforming to international / industry standards.

[0029] It should be noted that micro-point cloud data can accurately describe minute surface features (such as curvature, texture, and edges), while macro-point cloud data has a relatively low point cloud density and a wide coverage.

[0030] For example, the feature extraction based on calibration parameters includes: Acoustic signals are acquired, preprocessed, and feature maps are obtained by combining calibration parameters. Based on the feature maps, three types of feature parameters are calculated to obtain the first feature set. It should be added that the preprocessing of the acoustic signal is as follows: the acoustic signal is detrended, the Hanning window is applied to reduce spectral leakage, and the Daubechies 8th order wavelet is used to perform 3-level wavelet packet decomposition to obtain the feature map. The Daubechies 8th wavelet, with its core feature being the 8th order vanishing moment, is widely used in signal processing, image compression, denoising, and feature extraction. As it is an existing technology, this application will not elaborate further on it here.

[0031] Among them, the three types of characteristic parameters are: main frequency energy concentration. Time-domain waveform kurtosis and frequency domain entropy The calculation formulas are as follows: , , , In the formula: This represents the elastic modulus weighting coefficient, where E represents the measured elastic modulus. The center frequency of the k-th frequency component is... Represents frequency Spectral energy density at that location The total number of frequency components in the spectrum. This represents the set of defect-sensitive frequency bands. Represents the entire frequency band set. For frequency band indexing, This is the lowest frequency of frequency band b. This is the highest frequency of frequency band b. The magnitude coefficient of the exponential term. The decay rate of the exponential term. The amplitude coefficient of the sine term. Here, t is the frequency coefficient of the sinusoidal term, t is the time variable, and e is the natural constant; A second matrix is ​​constructed based on the concentration of the main frequency energy and the kurtosis of the time-domain waveform.

[0032] The step of fusing the registration data with the first feature set includes: Align the registration data with the first feature set, then map the first feature set onto the coordinate system of the registration data to obtain the first dataset and calculate the registration error. The calculation formula is as follows: ,

[0033] In the formula: ( , , ) represents the coordinates of the object on the X, Y, and Z axes obtained through acoustic measurements. , , () represents the coordinates of points on the object's surface obtained by laser scanning.

[0034] It should be noted that, The diameter must be ≤0.05mm to verify the spatial consistency between the acoustic scan and the point cloud data, and to ensure the accuracy of defect localization.

[0035] S103: Extract the abnormal regions from the first dataset to obtain the first matrix. Calculate the defect parameters using the first matrix, classify and quantify the defect parameters, and obtain the defect score. In this embodiment, the specific steps for extracting the abnormal region from the first dataset are as follows: First, calculate the curvature value of each point in the point cloud of the first dataset, sort the curvature values ​​from smallest to largest, and select the point with the smallest curvature as the initial seed point. By combining acoustic signals, the coordinate increments (Δx, Δy, Δz) of acoustic defects are aligned with the point cloud coordinates to locate potential anomaly areas; Growth criteria are defined based on attributes such as the angle between the normal vectors of neighboring points, curvature, and reflection intensity I. For example: Normal vector angle threshold: The angle between the normal vectors of the neighboring point and the seed point must be less than the smoothing threshold of 15° to ensure the geometric continuity of the region; The curvature of neighboring points must be less than 0.05 to avoid misclassifying high-curvature outliers into smooth regions; Starting from the seed point, traverse its neighboring points, adding points that satisfy the growth criteria to the current region and marking them as visited. Newly added points become new seed points, and growth continues outwards until no more points satisfy the criteria. Region growth terminates when the seed point queue is empty or no new points satisfy the growth criteria. At this point, the current region is saved as the clustering result, i.e., considered an outlier region.

[0036] The defect parameters are calculated using the first matrix as follows: Calculate the defect depth: In the formula: c is the speed of sound formula, and Δt is the round-trip time difference of the sound wave; The boundary extraction of point cloud anomaly regions is performed using a region growing algorithm, with the following formula: , In the formula: The coordinates of the points in the abnormal region; It should be noted that the classification and quantification of defect parameters includes: The physical characteristics of the defect parameters are verified, and the defects are classified and the severity score is calculated based on their geometric shape and the threshold of the characteristic parameters. like If ≥0.7 and K≥5, it is marked as a wrinkle defect. <0.7 and <2 is considered a hole or layered defect, and further refined according to the geometric shape of the defect. For example, (L / W>2 is judged as a long strip wrinkle, L / W≤2 is judged as a circular hole). The formula for calculating the severity score is as follows: , Based on defect type, size, coordinates, and severity score, a quantitative table is generated, as shown in the table below: Surface cracks 15.1×2.1×0.6mm (120,85,30) 0.81 Internal cavity Φ8.7mm (200,150,100) 0.62 Edge damage 6.0×3.0mm (0,50,0) 0.55 S104: Construct feature vectors based on the first dataset and defect parameters, enhance the feature vectors to form enhanced feature vectors, and identify defect areas.

[0037] Meanwhile, the feature vector Enhancement processing is performed, where H is the height dimension, W is the width dimension, C is the channel dimension, and F is the frequency domain dimension, including: A nonlinear topological mapping algorithm is used to reconstruct the eigenvectors, and topological invariants between dimensions are extracted using continuous cohomology theory to construct a topological feature matrix. Where β is the topological feature dimension, this matrix is ​​generated by the following formula: , In the formula: These are the topological weight coefficients. It is the sigmoid activation function. Let be the gradient field of the i-th topological field.

[0038] Align the reconstructed eigenvectors with each other to construct a Riemannian manifold M, whose metric tensor is derived from... Defined by the following formula: Let v be a vector field on the manifold. Using covariant derivatives, we obtain aligned multidimensional features; Fractal kernels are generated using the Mandelbrot set, and then convolutional operations are performed between the fractal kernels and aligned multi-dimensional features to form enhanced features. Among them, the Mandelbrot set is a set obtained in fractal geometry through complex number iteration formulas. The defined point set, where c is a complex constant. The complex value after the nth iteration; The enhanced feature vector is obtained by concatenating the enhanced feature and the feature vector and performing a Hadamard product operation.

[0039] In this embodiment, the specific steps for identifying the defective region are as follows: The enhanced feature vectors are mapped into three-dimensional space to obtain a three-dimensional feature point cloud. The fractal dimension of the neighborhood of each point is calculated. Based on the difference in fractal dimension, the defect region is segmented to obtain a defect region mask. A 3D mesh is constructed, the gradient field and normal vector of each mesh are calculated, and each mesh is marked by a defect region mask to obtain a 3D defect map. A transfer function is designed and used to visualize the 3D defect map to obtain the defect region.

[0040] The transfer function is:

[0041] In the formula: The opacity coefficient. The gradient decay coefficient is... The magnitude of the gradient reflects the rate of field change caused by the defect.

[0042] Example 2: Please see Figure 2 As shown, this embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements any of the control methods for pre-embedded blade end face wrinkles at the root of the wind turbine blade provided by the above methods.

[0043] Since the electronic device described in this embodiment is the electronic device used to implement the control method for pre-embedded blade end face wrinkles at the root of the wind turbine blade in this application embodiment, those skilled in the art can understand the specific implementation method and various variations of the electronic device in this embodiment based on the control method for pre-embedded blade end face wrinkles at the root of the wind turbine blade described in this application embodiment. Therefore, how the electronic device implements the method in this application embodiment will not be described in detail here. Any electronic device used by those skilled in the art to implement the control method for pre-embedded blade end face wrinkles at the root of the wind turbine blade in this application embodiment falls within the scope of protection of this application.

[0044] Example 3:

[0045] Please see Figure 3As shown, this embodiment discloses a computer-readable storage medium, including a memory, a processor, and a computer program stored on the memory and running on the processor. When the processor executes the computer program, it implements any of the control methods for pre-embedded blade end face wrinkles at the root of the wind turbine blade provided by the above methods.

[0046] 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, weights, and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0047] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired or wireless network. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0048] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this invention can be implemented in 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. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0049] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0050] In the several embodiments provided by this invention, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only one method, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0051] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0052] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0053] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

[0054] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for controlling the wrinkles on the end face of a pre-embedded blade at the root of a wind turbine blade, characterized in that, The method includes: S101: Preprocess the pre-embedded blade end face area at the root of the blade to obtain material properties, scan the preprocessed end face area and calculate the surface roughness of the end face area to obtain a DEM file; S102: The material properties are calibrated to obtain calibration parameters. Macroscopic point cloud data is acquired and aligned with the DEM file to form registration data. Feature extraction is performed based on the calibration parameters to obtain the first feature set. The registration data is fused with the first feature set to obtain the first dataset. S103: Extract the abnormal regions from the first dataset to obtain the first matrix. Calculate the defect parameters using the first matrix, classify and quantify the defect parameters, and obtain the defect score. S104: Construct feature vectors based on the first dataset and defect parameters, enhance the feature vectors to form enhanced feature vectors, and identify defect areas.

2. The method for controlling the end face wrinkles of the pre-embedded blade profile at the root of the wind turbine blade according to claim 1, characterized in that, The scanning of the pre-processed end face region and the calculation of the surface roughness of the end face region include: The scanning end face area generates micro-point cloud data, which is then optimized to obtain optimized point cloud data. The surface roughness of the cross-sectional area is then calculated based on the optimized point cloud data.

3. The method for controlling the wrinkles on the end face of the pre-embedded blade at the root of a wind turbine blade according to claim 2, characterized in that, The optimization of microview cloud data includes: The stiffness coefficient is calculated based on the material properties. The removal threshold is adjusted using the stiffness coefficient. The average distance of the neighboring points is calculated for each point. The average distance is compared with the adjusted removal threshold to obtain the denoised point cloud data. Calculate the local gradient of the denoised point cloud data, determine the interpolation weight coefficients based on the local gradient, and perform adaptive interpolation to obtain optimized point cloud data.

4. The method for controlling the end face wrinkles of the pre-embedded blade profile at the root of the wind turbine blade according to claim 1, characterized in that, The feature extraction based on calibration parameters includes: Acoustic signals are acquired, preprocessed, and feature maps are obtained by combining calibration parameters. Based on the feature maps, three types of feature parameters are calculated to obtain the first feature set. Among them, the three types of characteristic parameters are: main frequency energy concentration. Time-domain waveform kurtosis and frequency domain entropy The calculation formulas are as follows: ， ， ， In the formula: This represents the elastic modulus weighting coefficient, where E represents the measured elastic modulus. The center frequency of the k-th frequency component is... Represents frequency Spectral energy density at that location The total number of frequency components in the spectrum. This represents the set of defect-sensitive frequency bands. Represents the entire frequency band set. For frequency band indexing, This is the lowest frequency of frequency band b. This is the highest frequency of frequency band b. The magnitude coefficient of the exponential term. The decay rate of the exponential term. The amplitude coefficient of the sine term. Here, t is the frequency coefficient of the sinusoidal term, t is the time variable, and e is the natural constant; A second matrix is ​​constructed based on the concentration of the main frequency energy and the kurtosis of the time-domain waveform.

5. The method for controlling the end face wrinkles of the pre-embedded blade profile at the root of the wind turbine blade according to claim 1, characterized in that, The process of fusing the registration data with the first feature set includes: Align the registration data with the first feature set, and then map the first feature set onto the coordinate system of the registration data to obtain the first dataset and calculate the registration error.

6. The method for controlling the end face wrinkles of the pre-embedded blade profile at the root of the wind turbine blade according to claim 1, characterized in that, The classification and quantification of defect parameters includes: The physical characteristics of the defect parameters are verified, and the defects are classified and the severity score is calculated based on their geometric shape and the threshold of the characteristic parameters. A quantitative table is generated by integrating defect type, size, coordinates, and severity score.

7. The method for controlling the end face wrinkles of the pre-embedded blade profile at the root of the wind turbine blade according to claim 1, characterized in that, The enhancement process for the feature vector includes: The feature vectors are reconstructed, and the reconstructed feature vectors are aligned with each other to obtain aligned multi-dimensional features. Generate a fractal kernel, and perform a convolution operation between the fractal kernel and the aligned multi-dimensional features to form enhanced features; The enhanced feature vector is concatenated with the feature vector and then subjected to the Hadamard product operation to obtain the enhanced feature vector.

8. The method for controlling the end face wrinkles of the pre-embedded blade profile at the root of the wind turbine blade according to claim 1, characterized in that, The specific steps for identifying the defective region are as follows: The enhanced feature vectors are mapped into three-dimensional space to obtain a three-dimensional feature point cloud. The fractal dimension of the neighborhood of each point is calculated. Based on the difference in fractal dimension, the defect region is segmented to obtain a defect region mask. A 3D mesh is constructed, the gradient field and normal vector of each mesh are calculated, and each mesh is marked by a defect region mask to obtain a 3D defect map. A transfer function is designed and used to visualize the 3D defect map to obtain the defect region.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the control method for pre-embedded blade end face wrinkles at the root of the wind turbine blade according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed, implements the control method for pre-embedded blade end face wrinkles at the root of the wind turbine blade according to any one of claims 1 to 8.

Images