Fan blade damage correlation analysis method based on unmanned aerial vehicle inspection
By training the color abnormality and damage type identification model during drone inspection and conducting correlation analysis based on historical data, the problem of lack of in-depth root cause analysis of the blade inspection of fan blades in the existing technology is solved, providing a scientific decision-making basis, extending the service life of the blade and reducing replacement costs.
Patent Information
- Application Number
- CN202411880644.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-19
- Publication Date
- 2025-04-11
AI Technical Summary
The existing fan blade inspection methods can only simply identify the faults, and fail to deeply analyze the root causes of the faults, resulting in a lack of scientific basis for wind farm management and decision-making.
The fan blade damage correlation analysis method based on drone patrol is adopted. By training the color abnormality recognition model and the damage type recognition model, combining historical recognition results, recognition result time and fan data, the correlation analysis of abnormal area and damage recognition results is carried out, and the root cause analysis is carried out.
Provide quantitative data references to help managers and decision makers understand the causes of the failure more deeply, thereby taking more effective prevention and repair measures to extend the life of the blade and reduce replacement costs.
Smart Images
Figure CN119810695A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data processing, and in particular to a method for analyzing wind turbine blade damage based on unmanned aerial vehicle inspection. Background Art
[0002] The importance of drones in wind turbine blade inspection is self-evident. As the core component of wind turbines, wind turbine blades are exposed to harsh natural environments for a long time and are subjected to huge mechanical stress and environmental erosion, making blade inspection and maintenance an indispensable part of wind farm operations. The application of drone technology not only improves inspection efficiency and reduces safety risks, but also enables accurate identification and recording of blade surface defects. UAVs equipped with high-definition cameras and sensors can conduct detailed inspections of blades from a long distance and from multiple angles, greatly improving the accuracy and intelligence of inspections.
[0003] Although drone technology has made significant progress in the field of wind turbine blade inspection, the current wind turbine blade inspection often simply identifies whether there is a fault, without further analyzing the root cause of the fault. This inspection method lacks comprehensive analysis of multiple factors such as the surrounding environment and the blade's own factors, resulting in fault identification remaining on the surface and not going deep into the deep root cause of the fault. This is not conducive to the refined management and scientific decision-making of wind farms, because only by deeply understanding the cause of the fault can more effective prevention and repair measures be taken, thereby extending the service life of the blade and reducing replacement costs.
[0004] In view of this, a wind turbine blade damage correlation analysis method based on UAV inspection is needed. Summary of the invention
[0005] In view of the problem that the current inspection of wind turbine blades in the prior art often simply identifies whether there is a fault, but does not further analyze the root cause of the fault, the present invention provides a wind turbine blade damage correlation analysis method based on drone inspection, which can train a color anomaly recognition model and a damage type recognition model based on the wind turbine blade, and then obtain the abnormal area recognition result and the blade damage recognition result and the blade data based on the historical recognition results of the wind turbine blades and the time when the recognition results appear and the wind turbine data, respectively. The correlation between the results is sorted, and then the root cause analysis is performed based on this, providing quantitative data reference for managers and decision makers. The specific technical scheme is as follows:
[0006] A method for analyzing wind turbine blade damage based on drone inspection includes the following steps:
[0007] Obtain images of wind turbine blades by mounting a high-definition camera on a drone;
[0008] Training a color anomaly recognition model and a damage type recognition model based on fan blades; the color anomaly recognition model is used to identify the parts of the fan blades whose colors are different from the surrounding colors and to identify the areas of the parts; the damage type recognition model is used to identify whether there is a certain type of damage in the fan blade image;
[0009] Collect historical identification results of wind turbine blades, the time when the identification results appeared, and wind turbine data;
[0010] The abnormal area recognition result, the blade damage recognition result and the blade data are respectively obtained to sort the associations, and the first association sequence A=[a1, a2, ..., a i , …, a n ] and the second associated case sequence:
[0011] Obtain the overlapping data items in the first association situation sequence and the second association situation sequence, and sort them to construct a third association situation sequence C = [c1, c2, ..., c k , …, c u ].
[0012] Preferably, the training process of the color anomaly recognition model is as follows:
[0013] S1: Collect a large number of wind turbine blade images, including normal and stained or damaged blade images, and annotate the collected images, mark the areas with inconsistent colors, and then convert the images from RGB color space to Lab color space;
[0014] S2: Select a convolutional neural network as the base model, use CNN to extract features from the image, and train it to identify areas with inconsistent colors, then define a loss function and optimizer to train the model;
[0015] S03: Use OpenCV to convert the image from BGR color space to HSV or Lab color space; define the color range, use the `cv2.inRange()` function to extract the pixels within the color range, and create a mask; perform opening and closing operations on the mask to remove noise and connect broken areas;
[0016] S4: Use the `cv2.findContours()` function to extract the contour of the mask, then use the `cv2.drawContours()` function to draw the contour on the original image to frame the area with inconsistent colors, and finally use the `cv2.contourArea()` function to calculate the area of each contour;
[0017] S5: Output the results of the framed area and the corresponding color inconsistent area, and output the area of each color inconsistent area.
[0018] Preferably, the training process of the fan blade damage type recognition model is as follows:
[0019] S01: Collect a large number of wind turbine blade images, including normal images and images of blades with different types of damage (such as breakage, cracks, etc.); annotate the collected images, clearly indicating each type of damage. These annotations will be used as labels for model training; preprocess the images, including denoising and contrast enhancement.
[0020] S02: Select convolutional neural network as the basic model, use CNN to extract features from images, and train it to be able to identify different types of damage features; then define the loss function and optimizer to train the model;
[0021] S03: Divide the data into a training set and a validation set. Use the training set to train the model, and the validation set to verify and tune the model. Use the training set data to train the model, and adjust the model parameters through the validation set data to prevent overfitting.
[0022] S04: Use accuracy and recall metrics to evaluate model performance; adjust and optimize the model based on the evaluation results to improve the accuracy and robustness of damage detection.
[0023] Preferably, the process of sorting the degree of association between the blade damage identification results and the blade data is as follows:
[0024] Numerical encoding of all features in weather data and blade damage identification results;
[0025] Format the data into transaction datasets, where each transaction contains a set of feature items;
[0026] Calculate the support of the item, and then use the minimum support threshold Support(X) min , select the support greater than or equal to Support(X) min The calculation formula of the support of the item is as follows:
[0027]
[0028] Where X is an item set, N(X) is the number of transactions containing item set X, and N total is the total number of transactions;
[0029] For each frequent item set, a conditional pattern base and a conditional FP tree are constructed to mine deeper frequent item sets;
[0030] Use FP-tree to recursively mine frequent itemsets until no more frequent itemsets can be found;
[0031] Based on the frequent item sets, association rules are generated and the confidence of each rule is calculated. The calculation formula of the confidence is as follows:
[0032]
[0033] In the formula, X is the antecedent of the rule, Y is the consequent of the rule, Support(X∪Y) is the frequency of item sets X and Y appearing at the same time, and Support(X) is the frequency of item set X appearing;
[0034] Select a confidence level greater than or equal to the minimum confidence threshold min The rules are then sorted according to their confidence.
[0035] Preferably, the same method is used to obtain the association between the blade data and the blade color abnormality recognition result, and then the correlation degree ranking of the blade data on the blade damage recognition result is obtained, and finally the following is obtained:
[0036] The first association sequence of blade data to blade fault type identification result: A = [a1, a2, ..., a i , …, a n ]
[0037] The second association sequence of leaf data to leaf color abnormality recognition result: B = [b1, b2, ..., b j , …, b m ].
[0038] Preferably, the sorting method is: add the ranking of the item in the first association situation sequence to the ranking of the item in the second association situation sequence, and finally the items with higher rankings continue to be ranked higher in the third association situation sequence, and items with equal rankings have the same ranking in the third association situation sequence.
[0039] A computer-readable storage medium includes a stored program, wherein when the program is run, the device where the computer-readable storage medium is located is controlled to execute the wind blade damage correlation analysis method based on drone inspection as described above.
[0040] A processor is used to run a program, wherein when the program is run, the wind turbine blade damage correlation analysis method based on drone inspection as described above is executed.
[0041] Compared with the prior art, the present invention has the following beneficial effects:
[0042] The present invention obtains pictures of wind turbine blades by carrying a high-definition camera on a drone, and then trains a color anomaly recognition model and a damage type recognition model based on the wind turbine blades; the color anomaly recognition model is used to identify parts of the wind turbine blades whose colors are different from the surrounding colors and to identify the areas of the parts; the damage type recognition model is used to identify whether a certain type of damage exists in the pictures of the wind turbine blades, and then collects historical recognition results of the wind turbine blades, the time when the recognition results appear, and the wind turbine data, and finally obtains and sorts the correlation between the abnormal area recognition results and the blade damage recognition results and the blade data, and further obtains a comprehensive sorting to provide quantitative data reference for managers and decision makers. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following is a brief introduction to the drawings required for the specific embodiments or the description of the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn according to the actual scale.
[0044] Figure 1 is a flow chart of the method of the present invention;
[0045] Figure 2 Schematic diagram of the training process of the color anomaly recognition model;
[0046] Figure 3 Schematic diagram of the training process of the wind turbine blade damage type recognition model;
[0047] Figure 4 Schematic diagram of the process of sorting the degree of association. DETAILED DESCRIPTION
[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0049] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.
[0050] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.
[0051] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.
[0052] In one embodiment of the present invention, a method for analyzing wind turbine blade damage based on drone inspection is provided. Figure 1 As shown, the following steps are included:
[0053] Step 1: Obtain wind turbine blade images by mounting a high-definition camera on a drone;
[0054] Step 2: training a color anomaly recognition model and a damage type recognition model based on the fan blades; the color anomaly recognition model is used to identify the parts of the fan blades whose colors are different from the surrounding colors and to identify the areas of the parts; the damage type recognition model is used to identify whether there is a certain type of damage in the fan blade image;
[0055] like Figure 2 As shown, the training process of the color anomaly recognition model is as follows:
[0056] S1: Data collection and preprocessing, including:
[0057] Data collection: Collect a large number of wind turbine blade images, including normal and stained or damaged blade images.
[0058] Data annotation: Annotate the collected images and mark the areas with inconsistent colors. These areas will serve as targets for model training.
[0059] Preprocessing: Convert the image from RGB color space to Lab color space and apply Gaussian filtering to reduce noise.
[0060] S2: Model training: Convolutional neural network (CNN) is selected as the base model because CNN performs well in image recognition tasks. CNN is used to extract features from images and trained to identify areas with inconsistent colors. A loss function (such as cross entropy loss) and an optimizer (such as Adam optimizer) are defined to train the model.
[0061] S03: Color inconsistent area detection, including:
[0062] Color Space Conversion: Use OpenCV to convert the image from BGR color space to HSV or Lab color space as these color spaces are more suitable for color segmentation.
[0063] Color threshold segmentation: define the color range, use the `cv2.inRange()` function to extract pixels within the color range, and create a mask.
[0064] Morphological operations: Perform morphological operations on the mask, such as opening and closing operations, to remove noise and connect broken areas.
[0065] S4: Region marking and area calculation: Use the `cv2.findContours()` function to extract the contour of the mask. Then use the `cv2.drawContours()` function to draw contours on the original image to frame the areas with inconsistent colors. Finally, use the `cv2.contourArea()` function to calculate the area of each contour.
[0066] S5: Result output: Output the results of the framed area and the corresponding color inconsistent area, and output the area of each color inconsistent area.
[0067] S6: Model evaluation and optimization: Use indicators such as accuracy and recall to evaluate model performance. Adjust and optimize the model based on the evaluation results to improve the accuracy and robustness of detection.
[0068] S7: Deployment and Application: Deploy the trained model to real-time inspection of wind turbine blades. As new data is collected, the model is retrained to adapt to new stain types.
[0069] Through the above steps, we can build a model that can identify color-inconsistent regions on wind turbine blades and calculate their areas, and apply it to actual detection tasks.
[0070] like Figure 3 As shown, the training process of the fan blade damage type recognition model is as follows:
[0071] S01: Data collection and preprocessing, including the following steps:
[0072] Data collection: Collect a large number of wind turbine blade images, including normal blades and blades with different types of damage (such as breakage, cracks, etc.).
[0073] Data annotation: Annotate the collected images to clearly indicate each type of damage. These annotations will serve as labels for model training.
[0074] Preprocessing: Preprocess the image, including denoising, contrast enhancement, and adjusting brightness and contrast to highlight damaged features.
[0075] S02: Model selection and training: Convolutional neural network (CNN) is selected as the base model because CNN performs well in image recognition tasks, especially for problems such as damage detection. CNN is used to extract features from images and trained to recognize different types of damage features. Then a loss function (such as cross entropy loss) and an optimizer (such as Adam optimizer) are defined to train the model.
[0076] S03: Model training and validation: Divide the data into training set and validation set. Use the training set to train the model, and the validation set to verify and tune the model. Use the training set data to train the model, and adjust the model parameters through the validation set data to prevent overfitting.
[0077] S04: Model evaluation and optimization: Evaluate model performance using metrics such as accuracy and recall. Adjust and optimize the model based on the evaluation results to improve the accuracy and robustness of damage detection.
[0078] S05: Deployment and Application: Deploy the trained model to real-time damage detection on wind turbine blades. As new data is collected, retrain the model to adapt to new damage types and improve detection accuracy.
[0079] S6: Based on the trained damage type recognition model, the wind turbine blade images are identified and the detection results, including damage type and location, are displayed to provide decision support for maintenance and repair.
[0080] Through the above steps, we can build a model that can identify different types of damage on wind turbine blades and apply it to the actual damage detection task.
[0081] Step three: Collect historical identification results of wind turbine blades and the time when the identification results appear. The identification results include: abnormal area identification results and blade damage identification results. The abnormal area identification results include: abnormal blade color, abnormal area, and normal blade color; the blade damage identification results include: no damage to the blade, damaged blade, and damage type. Collect blade data of different blade numbers at different times. The blade data include: blade service life, manufacturing materials, manufacturer, purchase price, ambient temperature and humidity, geographical location altitude, temperature difference between morning and evening, rainfall, and ice coverage duration.
[0082] Step 4: Obtain the association between the abnormal area recognition result and the blade damage recognition result and the blade data respectively, and rank the degree of association between the recognition result and the blade data. Since the abnormal area is obtained in the same way as the association between the abnormal area recognition result and the blade damage recognition result and the blade data, the following takes the blade damage recognition result as an example. Figure 4As shown, the process of sorting the degree of association between the blade damage identification results and the blade data is given:
[0083] S401: Numerically encode all features in the weather data (such as blade age, manufacturing material, manufacturer, purchase price, ambient temperature and humidity, geographical location altitude, temperature difference between morning and evening, rainfall, and ice-covering duration) and blade damage identification results so that the algorithm can process them;
[0084] S402: Format the data into a transaction data set, where each transaction contains a set of feature items. For example, a transaction can be "high blade age, medium manufacturing material, high purchase price, high ambient temperature and humidity, high geographical location altitude, high temperature difference between morning and evening, low rainfall, and low ice cover duration", and add a "blade damage identification result" label at the end of the transaction.
[0085] S403: Construct an FP-tree. The FP-tree is a compressed data structure for storing frequent item sets. The construction steps are: calculate the support of the item, and then according to the minimum support threshold Support(X) min , select the support greater than or equal to Support(X) min The calculation formula of the support of the item is as follows:
[0086]
[0087] Where X is an item set, N(X) is the number of transactions containing item set X, and N total is the total number of transactions;
[0088] S404: For each frequent item set, a conditional pattern base (CM) and a conditional FP (CFPT) tree are constructed to mine deeper frequent item sets.
[0089] S405: recursively mine frequent item sets using the FP-tree until no more frequent item sets can be found;
[0090] S406: Generate association rules based on frequent item sets and calculate the confidence of each rule. The calculation formula of the confidence is as follows:
[0091]
[0092] In the formula, X is the antecedent of the rule, Y is the consequent of the rule, Support(X∪Y) is the frequency of item sets X and Y appearing at the same time, and Support(X) is the frequency of item set X appearing.
[0093] S407: Select a confidence level greater than or equal to the minimum confidence threshold Confidence min Then the rules are sorted according to their confidence.
[0094] After the above steps, several data items that affect the blade damage identification results can be determined, as well as the order of these data items based on the degree of influence. In other words, it can be determined which data in the blade data has the greatest influence on the blade damage identification results, which data has the second greatest influence on the blade damage identification results, and so on. At this point, the analysis of the influence of blade data on blade damage identification results is completed.
[0095] Afterwards, the association between the blade data and the blade color anomaly recognition result is obtained, and then the correlation degree ranking of the blade data on the blade damage recognition result is obtained.
[0096] In summary, this step can be concluded as follows:
[0097] The first association sequence of blade data to blade fault type identification result: A = [a1, a2, ..., a i , …, a n ]
[0098] The second association sequence of leaf data to leaf color anomaly recognition results:
[0099] Step 5: Obtain the overlapping data items in the first association situation sequence and the second association situation sequence, and sort them to construct a third association situation sequence C = [c1, c2, ..., c k , …, c u ]
[0100] In this step, the sorting method is: add the ranking of the item in the first association situation sequence to the ranking of the item in the second association situation sequence, and finally the items with higher rankings continue to be ranked higher in the third association situation sequence, and the items with equal rankings have the same ranking in the third association situation sequence.
[0101] In summary, the present invention obtains pictures of wind turbine blades by carrying a high-definition camera on a drone, and then trains a color anomaly recognition model and a damage type recognition model based on the wind turbine blades; the color anomaly recognition model is used to identify parts of the wind turbine blades whose colors are different from the surrounding colors and to identify the areas of the parts; the damage type recognition model is used to identify whether there is a certain type of damage in the wind turbine blade pictures, and then collects the historical recognition results of the wind turbine blades, the time when the recognition results appear, and the wind turbine data, and finally obtains the correlation between the abnormal area recognition results and the blade damage recognition results and the blade data, and further obtains a comprehensive ranking to provide quantitative data reference for managers and decision makers.
[0102] Those of ordinary skill in the art will appreciate that the units of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition of each example has been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.
[0103] In the embodiments provided by the present invention, it should be understood that the division of units is only a logical function division, and there may be other division methods in actual implementation, for example, multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored, etc.
[0104] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.
[0105] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-0nlyMemory), random access memory (RAM, RandomAccessMemory), mobile hard disk, magnetic disk or optical disk, etc., which can store program code.
[0106] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. These modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the claims and specification of the present invention.
Claims
1. A wind turbine blade damage correlation analysis method based on drone inspection, characterized in that: The following steps are involved: Obtain images of wind turbine blades by mounting a high-definition camera on a drone; Training a color anomaly recognition model and a damage type recognition model based on fan blades; the color anomaly recognition model is used to identify the parts of the fan blades whose colors are different from the surrounding colors and to identify the areas of the parts; the damage type recognition model is used to identify whether there is a certain type of damage in the fan blade image; Collect historical identification results of wind turbine blades, the time when the identification results appeared, and wind turbine data; The abnormal area recognition result, the blade damage recognition result and the blade data are respectively obtained to sort the associations, and the first association sequence A=[a1, a2, ..., a i , …, a n ] and the second associated case sequence: Obtain the overlapping data items in the first association situation sequence and the second association situation sequence, and sort them to construct a third association situation sequence C = [c1, c2, ..., c k , …, c u ].
2. The method for analyzing wind turbine blade damage based on drone inspection according to claim 1 is characterized in that: The training process of the color anomaly recognition model is as follows: S1: Collect a large number of wind turbine blade images, including normal and stained or damaged blade images, and annotate the collected images, mark the areas with inconsistent colors, and then convert the images from RGB color space to Lab color space; S2: Select a convolutional neural network as the base model, use CNN to extract features from the image, and train it to identify areas with inconsistent colors, then define a loss function and optimizer to train the model; S3: Use OpenCV to convert the image from BGR color space to HSV or Lab color space; define the color range, use the `cv2.inRange()` function to extract the pixels within the color range, and create a mask; perform opening and closing operations on the mask to remove noise and connect broken areas; S4: Use the `cv2.findContours()` function to extract the contour of the mask, then use the `cv2.drawContours()` function to draw the contour on the original image to frame the area with inconsistent colors, and finally use the `cv2.contourArea()` function to calculate the area of each contour; S5: Output the results of the framed area and the corresponding color inconsistent area, and output the area of each color inconsistent area.
3. The method for analyzing wind turbine blade damage based on drone inspection according to claim 1 is characterized in that: The training process of the fan blade damage type recognition model is as follows: S01: Collect a large number of wind turbine blade images, including normal blades and blades with different types of damage; annotate the collected images, clearly indicating each type of damage, and these annotations will be used as labels for model training; preprocess the images, including denoising and contrast enhancement; S02: Select convolutional neural network as the basic model, use CNN to extract features from images, and train it to be able to identify different types of damage features; then define the loss function and optimizer to train the model; S03: Divide the data into a training set and a validation set. Use the training set to train the model, and the validation set to verify and tune the model. In the process of using the training set data to train the model, adjust the model parameters through the validation set data to prevent overfitting. S04: Use accuracy and recall metrics to evaluate model performance; adjust and optimize the model based on the evaluation results to improve the accuracy and robustness of damage detection.
4. The method for analyzing wind turbine blade damage based on drone inspection according to claim 1 is characterized in that: The process of sorting the degree of correlation between blade damage identification results and blade data: Numerical encoding of all features in weather data and blade damage identification results; Format the data into transaction datasets, where each transaction contains a set of feature items; Calculate the support of the item, and then use the minimum support threshold Support(X) min , select the support greater than or equal to Support(X) min The calculation formula of the support of the item is as follows: Where X is an item set, N(X) is the number of transactions containing item set X, and N total is the total number of transactions; For each frequent item set, a conditional pattern base and a conditional FP tree are constructed to mine deeper frequent item sets; Use FP-tree to recursively mine frequent itemsets until no more frequent itemsets can be found; Based on the frequent item sets, association rules are generated and the confidence of each rule is calculated. The calculation formula of the confidence is as follows: In the formula, X is the antecedent of the rule, Y is the consequent of the rule, Support(X∪Y) is the frequency of item sets X and Y appearing at the same time, and Support(X) is the frequency of item set X appearing; Select a confidence level greater than or equal to the minimum confidence threshold min The rules are then sorted according to their confidence.
5. The method for analyzing wind turbine blade damage based on drone inspection according to claim 4 is characterized in that: The same method is used to obtain the correlation between the blade data and the blade color abnormality recognition results, and then the correlation degree ranking of the blade data on the blade damage recognition results is obtained, and finally the following is obtained: The first association sequence of blade data to blade fault type identification result: A = [a1, a2, ..., a i , …, a n ] The second association sequence of leaf data to leaf color anomaly recognition results:
6. The method for analyzing wind turbine blade damage based on drone inspection according to claim 1 is characterized in that: The sorting method is: sort by the combined sum of the ranking in the first association situation sequence and the ranking in the second association situation sequence. Finally, the ones with higher rankings will continue to be ranked higher in the third association situation sequence, and the ones with equal rankings will have the same ranking in the third association situation sequence.
7. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute the wind blade damage correlation analysis method based on drone inspection as described in any one of claims 1 to 6.
8. A processor, characterized in that: The processor is used to run a program, wherein the program, when running, executes the wind blade damage correlation analysis method based on drone inspection as described in any one of claims 1 to 6.
Citation Information
Cited By
Wind turbine generator blade fracture damage detection method applying inspection robot
CN120889713A