An image-based fan blade aging detection method

Abstract

This invention provides an image-based method for detecting the aging of wind turbine blades, belonging to the field of image processing technology. In this invention, external images and internal ultrasonic images of wind turbine blades are acquired. By combining the aging conditions on the exterior and the internal aging conditions, the aging assessment value of the wind turbine blades is obtained. This invention solves the problem of the lack of existing methods for detecting the aging of wind turbine blades.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and more specifically, to an image-based method for detecting the aging of wind turbine blades. Background Technology

[0002] A single 1500kW wind turbine occupies approximately 17*17m, or 289 square meters. The spacing between turbines is typically 500m, meaning a 10,000kW wind turbine installation occupies approximately 2300 acres. Whether considering the footprint of a single turbine or the entire wind farm, manually inspecting the turbine blades is extremely difficult. After installation, it's challenging to monitor the aging of the blades. Most existing wind turbine blades are typically between 50m and 70m in length; therefore, if they age and detach, it poses a significant threat to the turbine itself or the safety of personnel at the wind farm. Summary of the Invention

[0003] In view of the above-mentioned shortcomings in the prior art, the present invention provides an image-based wind turbine blade aging detection method, which solves the problem of the lack of existing methods for detecting wind turbine blade aging.

[0004] To achieve the above-mentioned objectives, the technical solution adopted by this invention is: an image-based method for detecting the aging of wind turbine blades, comprising:

[0005] S1. Acquire images of the wind turbine blades using a camera;

[0006] S2. Acquire ultrasonic images of the inside of the wind turbine blades using an ultrasonic detector;

[0007] S3. Based on the appearance image, obtain the aging degree value of the fan blades;

[0008] S4. Based on the ultrasonic images, obtain the aging degree value inside the fan blades;

[0009] S5. Based on the aging degree values ​​of the wind turbine blades' appearance and internal aging degree values, obtain the aging assessment value of the wind turbine blades.

[0010] Furthermore, step S3 includes the following sub-steps:

[0011] S31. Classify each pixel in the appearance image to obtain the pixel distribution area of ​​each aging stage.

[0012] S32. Calculate the aging degree value of the fan blade appearance based on the pixel distribution area of ​​each aging stage.

[0013] Furthermore, step S31 includes the following sub-steps:

[0014] S311. Set the aging threshold sequence for each aging stage {R} j,thG j,th B j,th}, where R j,th G is the threshold for the R channel in the j-th aging stage. j,th For the G channel threshold of the j-th aging stage, B j,th Let K be the threshold of channel B for the j-th aging stage, where j = 1, ..., K, and K is the number of aging stages.

[0015] S312. Calculate the distance between each pixel on the appearance image and the channel threshold in each aging stage to obtain the first aging distance;

[0016] S313. Find the aging stage corresponding to the smallest distance among multiple first aging distances of the same pixel, and classify the pixel into that aging stage.

[0017] S314. All pixels in the same aging stage constitute the pixel distribution area of ​​that aging stage.

[0018] The beneficial effects of the above-mentioned further solution are as follows: The present invention sets channel thresholds for each aging stage, and determines the aging stage to which each pixel belongs based on the distance between each pixel on the acquired appearance image and the set channel thresholds.

[0019] Furthermore, the formula for calculating the first aging distance in S312 is as follows:

[0020] d i，j =(R i -R j，th ) 2 +(G i -G j，th ) 2 +(B i -B j，th ) 2

[0021] Where, d i，j R is the first aging distance between the i-th pixel in the appearance image and the j-th aging stage. i G is the R channel value of the i-th pixel in the appearance image. i Let B be the G channel value of the i-th pixel in the appearance image. i Let be the B channel value of the i-th pixel in the appearance image.

[0022] Furthermore, the formula for calculating the degree of aging of the wind turbine blades in step S32 is as follows:

[0023]

[0024] Where O1 represents the degree of aging of the wind turbine blades, γ lLet x be the aging weight of the pixel distribution area in the l-th aging stage, where L is the number of pixel distribution areas in the aging stage, and x is the aging weight of the pixel distribution area in the l-th aging stage. l This represents the number of pixels in the pixel distribution area during the l-th aging stage.

[0025] The beneficial effect of the above-mentioned further scheme is that it weights the distribution of pixels in various distribution areas and assigns different aging weights to various distributions, thereby realizing the assessment of the aging degree of the overall appearance.

[0026] Furthermore, step S4 includes the following sub-steps:

[0027] S41. Use the YOLO neural network to extract the crack region from the ultrasound image;

[0028] S42. Set the crack threshold sequence for each aging stage {E} j}, j = 1, ..., K, where K is the number of aging stages, and E j Let be the crack size threshold for the j-th aging stage;

[0029] S43. Calculate the distance between the crack region and the crack size threshold of each aging stage to obtain the second aging distance;

[0030] S44. Based on the aging stage corresponding to the minimum second aging distance, obtain the aging degree value inside the wind turbine blade.

[0031] Furthermore, the formula for calculating the second aging distance in S43 is as follows:

[0032] D j =(EE) j ) 2

[0033] Among them, D j E represents the distance between the crack region and the crack size threshold of the j-th aging stage. j Let E be the crack size threshold for the j-th aging stage, and E be the size of the crack region.

[0034] Furthermore, the formula for calculating the aging assessment value of the wind turbine blades in S5 is as follows:

[0035] O=ω1O1+ω2O2

[0036] Where O is the aging assessment value of the wind turbine blade, O1 is the aging degree value of the wind turbine blade appearance, O2 is the aging degree value of the wind turbine blade interior, ω1 is the weight value of appearance aging, and ω2 is the weight value of interior aging.

[0037] The technical solution of the present invention has at least the following advantages and beneficial effects: The present invention collects external images and internal ultrasonic images of wind turbine blades, and obtains the aging assessment value of wind turbine blades by comprehensively considering the aging conditions of the external appearance and the internal aging conditions. Attached Figure Description

[0038] Figure 1 This is a flowchart of an image-based method for detecting the aging of wind turbine blades. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0040] like Figure 1 As shown, an image-based method for detecting the aging of wind turbine blades includes:

[0041] S1. Acquire images of the wind turbine blades using a camera;

[0042] S2. Acquire ultrasonic images of the inside of the wind turbine blades using an ultrasonic detector;

[0043] S3. Based on the appearance image, obtain the aging degree value of the fan blades;

[0044] S3 includes the following steps:

[0045] S31. Classify each pixel in the appearance image to obtain the pixel distribution area of ​​each aging stage.

[0046] S31 includes the following steps:

[0047] S311. Set the aging threshold sequence for each aging stage {R} j，th G j，th B j，th}, where R j，th G is the threshold for the R channel in the j-th aging stage. j，th For the G channel threshold of the j-th aging stage, E j，th Let be the B-channel threshold for the j-th aging stage, where j = 1, ..., K, and K is the number of aging stages;

[0048] S312. Calculate the distance between each pixel on the appearance image and the channel threshold in each aging stage to obtain the first aging distance;

[0049] The formula for calculating the first aging distance in S312 is as follows:

[0050] d i，j =(R i -R j，th ) 2 +(G i -G j，th ) 2 +(B i -B j，th ) 2

[0051] Where, d i，j R is the first aging distance between the i-th pixel in the appearance image and the j-th aging stage. i G is the R channel value of the i-th pixel in the appearance image. i Let B be the G channel value of the i-th pixel in the appearance image. i Let be the B channel value of the i-th pixel in the appearance image;

[0052] S313. Find the aging stage corresponding to the smallest distance among multiple first aging distances of the same pixel, and classify the pixel into that aging stage.

[0053] S314. All pixels in the same aging stage constitute the pixel distribution area of ​​that aging stage.

[0054] S32. Calculate the aging degree value of the fan blade appearance based on the pixel distribution area of ​​each aging stage.

[0055] The formula for calculating the appearance aging degree of the wind turbine blades in S32 is as follows:

[0056]

[0057] Where O1 represents the degree of aging of the wind turbine blades, γ l Let x be the aging weight of the pixel distribution area in the l-th aging stage, where L is the number of pixel distribution areas in the aging stage, and x is the aging weight of the pixel distribution area in the l-th aging stage. l This represents the number of pixels in the pixel distribution area during the l-th aging stage.

[0058] S4. Based on the ultrasonic images, obtain the aging degree value inside the fan blades;

[0059] S4 includes the following sub-steps:

[0060] S41. Use the YOLO neural network to extract the crack region from the ultrasound image;

[0061] S42. Set the crack threshold sequence for each aging stage {E} j}, j = 1, ..., K, where K is the number of aging stages, and E j Let be the crack size threshold for the j-th aging stage;

[0062] S43. Calculate the distance between the crack region and the crack size threshold of each aging stage to obtain the second aging distance;

[0063] The formula for calculating the second aging distance in S43 is as follows:

[0064] D j =(EE) j ) 2

[0065] Among them, D j E represents the distance between the crack region and the crack size threshold of the j-th aging stage. j Let E be the crack size threshold for the j-th aging stage, and E be the size of the crack region.

[0066] S44. Based on the aging stage corresponding to the minimum second aging distance, obtain the aging degree value inside the wind turbine blade.

[0067] During the S4 process, the internal aging degree values ​​of the wind turbine blades corresponding to each aging stage are determined in advance based on experience.

[0068] S5. Based on the aging degree values ​​of the wind turbine blades' appearance and internal aging degree values, obtain the aging assessment value of the wind turbine blades.

[0069] The formula for calculating the aging assessment value of the wind turbine blades in S5 is as follows:

[0070] O=ω1O1+ω2O2

[0071] Where O is the aging assessment value of the wind turbine blade, O1 is the aging degree value of the wind turbine blade appearance, O2 is the aging degree value of the wind turbine blade interior, ω1 is the weight value of appearance aging, and ω2 is the weight value of interior aging.

[0072] The technical solution of the present invention has at least the following advantages and beneficial effects: The present invention collects external images and internal ultrasonic images of wind turbine blades, and obtains the aging assessment value of wind turbine blades by comprehensively considering the aging conditions of the external appearance and the internal aging conditions.

[0073] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. An image-based fan blade aging detection method, characterized in that, The method comprises the following steps: S1, acquiring an appearance image of a fan blade by a camera; S2, acquiring an internal ultrasonic image of the fan blade by an ultrasonic detector; S3, obtaining an appearance aging degree value of the fan blade according to the appearance image; S31, classifying each pixel point on the appearance image to obtain a pixel point distribution area of each aging stage; S32, calculating the appearance aging degree value of the fan blade according to the pixel point distribution area of each aging stage; The formula for calculating the appearance aging degree value of the fan blade in S32 is: Wherein, O1 is the fan blade appearance aging degree value, γ l is the aging weight of the pixel point distribution area of the lth aging stage, L is the number of pixel point distribution areas of the aging stage, x l is the number of pixel points on the pixel point distribution area of the lth aging stage; S4, obtaining an internal aging degree value of the fan blade according to the internal ultrasonic image; S41, extracting a crack area of the internal ultrasonic image by using a YOLO neural network; S42, set the crack threshold sequence {E j} of each aging stage, j = 1,..., K, K is the number of aging stages, wherein, E j is the crack size threshold of the jth aging stage; S43, calculating a second aging distance by calculating the distance between the crack area and the crack size threshold of each aging stage; S44, obtaining the internal aging degree value of the fan blade according to the aging stage corresponding to the minimum second aging distance; S5, obtaining a fan blade aging evaluation value according to the appearance aging degree value and the internal aging degree value of the fan blade; The formula for calculating the fan blade aging evaluation value in S5 is: Wherein, O is the fan blade aging evaluation value, O1 is the appearance aging degree value of the fan blade, O2 is the internal aging degree value of the fan blade, ω1 is the appearance aging weight value, and ω2 is the internal aging weight value.

2. The image-based fan blade aging detection method of claim 1, wherein, The S31 comprises the following steps: S311, setting an aging threshold sequence of each aging stage wherein, Rth(j) is the R channel threshold of the jth aging stage, Gth(j) is the G channel threshold of the jth aging stage, Bth(j) is the B channel threshold of the jth aging stage, j = 1,..., K, K is the number of aging stages; S312, calculating the distance between each pixel point on the appearance image and the channel threshold in each aging stage to obtain a first aging distance; S313, finding the aging stage corresponding to the minimum distance in the multiple first aging distances of the same pixel point, and classifying the pixel point into the aging stage; S314, forming a pixel point distribution area of the aging stage by all pixel points in the same aging stage.

3. The image-based fan blade aging detection method of claim 2, wherein, The formula for calculating the first aging distance in S312 is: wherein, Ri,j is the first aging distance of the i-th pixel point on the appearance image and the j-th aging stage, R i Gi is the R channel value of the i-th pixel point on the appearance image, G i Gi is the R channel value of the i-th pixel point on the appearance image, G i Gi is the R channel value of the i-th pixel point on the appearance image, G 4. The image-based fan blade aging detection method of claim 1, wherein, The formula for calculating the second aging distance in S43 is: Wherein, Dj is the distance between the crack area and the crack size threshold of the jth aging stage, Ej is the crack size threshold of the jth aging stage, and E is the size of the crack area.

Images