Method for identifying blade shape and clearance based on multiple identification points of blade

Abstract

The present application relates to the field of wind turbine blade shape and clearance monitoring technology, and particularly relates to a video identification blade shape and clearance method based on multiple identification points of a blade. A camera is arranged near the blade of a cabin, n identification points are arranged on the blade, a video of the blade passing through a tower drum is shot, the plane positions of the identification points are identified and calculated according to the video information, a blade shape curve is fitted and a load condition is calculated, the tip position is calculated according to the blade shape curve, and a clearance value is calculated. According to the method, the multiple identification points are arranged on the blade, the blade shape curve is fitted according to the position information of the identification points and the blade root, and the shape change condition and the load condition of the blade are obtained; meanwhile, the clearance value is obtained through the blade shape curve, and the clearance value is more accurate and reliable.

Description

Technical Field

[0001] This invention relates to the field of wind turbine blade morphology and clearance monitoring technology, specifically to a video recognition method for blade morphology and clearance based on multiple marker points on the blade. Background Technology

[0002] In pursuit of lower cost per kilowatt-hour, larger wind turbines have become the mainstream trend in the wind power industry. Increased turbine capacity, larger models, and longer blades have brought more challenges to the safe and stable operation of the units, such as the risk of turbine sweeping. To ensure the safe operation of the units, the number of airspace monitoring systems and blade deformation monitoring systems for wind turbines is increasing. Currently, commonly used airspace monitoring systems include image-based systems for blade tip identification, distance sensor-based laser airspace monitoring systems, and millimeter-wave radar airspace monitoring systems. Blade deformation monitoring includes image-based methods such as internal blade images and blade root images, as well as distance sensor-based methods such as blade tip displacement identification, internal blade tension wires, and blade root strain gauges.

[0003] For example, Chinese patent CN116816613A discloses a wind turbine blade morphology monitoring system. This system uses a first, second, and third image capture device installed on the outside of the hub cover. These image capture devices rotate with the wind turbine while remaining relatively stationary with the blades. This allows for the acquisition of blade morphology, hub speed, and blade pitch angle parameters during wind turbine operation through image recognition. This enables monitoring of the wind turbine load, blade clearance, and tower clearance, ensuring safe operation. However, the image capture devices are installed on the outside of the hub cover, requiring real-time acquisition of hub speed and blade pitch angle information. Furthermore, blade deformation at this angle may obscure the blade tip in the image, making it impossible to identify blade morphology and understand morphological changes under partial load conditions. Chinese patent CN117028161A discloses a wind turbine blade clearance detection system, method, equipment, and storage medium, including a deformation detection unit and a clearance detection unit. However, its deformation detection unit can only detect blade root deformation data and cannot identify the entire blade morphology. Summary of the Invention

[0004] This invention provides a video recognition method for blade morphology and clearance based on multiple marker points on the blade. This method arranges multiple marker points on the blade, fits the blade morphology curve according to the position information of each marker point and the blade root, and obtains the morphological changes and load conditions of the blade. At the same time, the clearance value is obtained through the blade morphology curve, and the clearance value is more accurate and reliable.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] A video-based method for identifying blade morphology and clearance based on multiple blade markers includes the following steps:

[0007] S1. Deploy cameras near the blades in the nacelle. Arrange markers at n+1 equal division points between the blade root and the blade tip. Arrange n markers on each blade, and all markers are located on the leeward side of the blade. Record the distance r between each marker and the blade root in sequence.

[0008] S2. Capture video of the blades passing through the tower, and identify and calculate the minimum distance d between each marker point and the tower based on the video information;

[0009] S3. Calculate the planar position of each marker point based on the distance r between each marker point and the leaf root and the minimum distance d between the corresponding marker point and the tower.

[0010] S4. Based on the planar position of each marker point and the blade model, fit the blade morphology curve and calculate the load.

[0011] S5. Calculate the leaf tip position based on the leaf morphology curve;

[0012] S6. Calculate the headroom value based on the blade tip position and fan structural parameters.

[0013] Each leaf has no fewer than five markings.

[0014] In S1, the distance between each marker point and the leaf root is calculated based on the leaf length.

[0015] The specific method for calculating the minimum distance d between each marker point and the tower in S2 is as follows:

[0016] S2.1 After the camera is installed, calibrate the conversion ratio between pixel distance in the video image and actual object distance;

[0017] S2.2. Based on the video information, identify the pixel position of the marker point on the blade in each frame of the image when the blade passes the tower, fit the curve to obtain the pixel position of each marker point closest to the tower, and identify the pixel position of the tower in the horizontal direction when each marker point is closest to the tower, so as to obtain the minimum pixel distance between each marker point and the tower.

[0018] S2.3. Based on the conversion ratio of pixel distance to actual distance, convert the minimum pixel distance between each marker point and the tower into the minimum distance d between each marker point and the tower.

[0019] The specific calculation method for the planar position of each marker point in S3 is as follows:

[0020] S3.1. With the blade root position as the origin 0, establish the x-axis coordinate along the horizontal direction from the nose to the tail of the nacelle, and establish the y-axis coordinate along the vertical direction.

[0021] S3.2. Based on the wind turbine structural parameters, the abscissa of the point on the edge of the tower near the blade is t0.

[0022] S3.3, Assume the coordinates of the i-th marker are (x... i y i Let i be an integer from 1 to n, and let r be the distance between the i-th marker and the leaf root. i Then x i 2 +y i 2 =r i 2 ;

[0023] S3.4, Given that the minimum distance between the i-th marker and the tower is d i Then x i =t0-d i ;

[0024] S3.5, The coordinates of the i-th marker point are calculated as follows: This allows us to obtain the coordinates of each marker point.

[0025] The methods for fitting leaf morphology curves in S4 include linear regression, multinomial regression, locally weighted scatter smoothing, logistic regression, and nonlinear regression.

[0026] Based on the blade morphology and vibration modes, a polynomial regression method is used to fit the blade morphology curve. The specific steps are as follows:

[0027] S4.1 Using the third-order polynomial regression method, then y = w0 + w1x + w2x 2 +w3x 3 ;

[0028] S4.2 Given the leaf root position and the planar positions of n marker points, the optimal parameters w0, w1, w2, and w3 are obtained using the method of minimizing the error of scattered points, thus obtaining the leaf morphology curve.

[0029] The specific calculation method for the load in S4 is as follows: After obtaining the blade shape curve, the blade shape curve is compared with the shape of the blade in the non-stressed state to obtain the deformation of each point, and then the current shape load of the blade is obtained.

[0030] In S5, the position of the leaf tip is calculated based on the distance between the leaf tip and the leaf root and the fitted leaf morphology curve.

[0031] The airspace value and blade morphology changes are fed back to the main controller, which then determines whether an early warning is needed and activates the airspace protection strategy.

[0032] Compared with the prior art, the beneficial effects of the present invention are: (1) In the present invention, by arranging multiple marker points on the blade, the minimum distance between each marker point and the tower when the blade passes through the tower is identified by video image, so as to obtain the planar position of each marker point when the blade is closest to the tower. Then, according to the planar position of each marker point, the position information of the blade root and the wind turbine structural parameters, the blade shape curve when the blade is closest to the tower is obtained. By comparing the blade shape curve with the shape of the blade in the non-stressed state, the shape change of the blade is obtained, and then the current shape load of the blade is obtained. At the same time, the blade tip position and the clearance value are calculated through the blade shape curve.

[0033] (2) In this invention, the clearance value is obtained by identifying the location of multiple marker points, fitting the blade morphology, and then calculating it, making the clearance value more accurate and reliable.

[0034] (3) In this invention, the main controller can issue an early warning and activate the airspace protection strategy based on the airspace value, load conditions and deformation conditions to protect the safety of the unit. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the video recognition method for blade morphology and clearance based on multiple blade markers according to the present invention;

[0036] Figure 2 This is a schematic diagram of the video image trajectory synthesis when the blade passes through the tower in this invention;

[0037] Figure 3 This is a schematic diagram of the coordinate system established with the leaf root as the origin in this invention;

[0038] In the image: 1-leaf root, 2-leaf tip, 3-camera, 4-marker point, 5-fitted leaf morphology curve, 6-tower. Detailed Implementation

[0039] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0040] This invention provides a video-based method for identifying blade morphology and clearance based on multiple marker points 4 on the blade, such as... Figure 1 As shown, it includes the following steps:

[0041] S1. Deploy camera 3 near the blade in the nacelle. Arrange marker points 4 at n+1 equally spaced points between the blade root 1 and blade tip 2. Arrange n marker points 4 on each blade, with all marker points located on the leeward side of the blade, i.e., facing the tower during normal power generation (approximately 0 degrees pitch angle). Record the distance r between each marker point 4 and the blade root 1 sequentially, i.e., the distances between each marker point 4 and the blade root 1 are r1, r2, ..., r... n ;

[0042] Specifically, the distance between each marker point 4 and the leaf root 1 can be calculated based on the leaf length. Given that n marker points 4 are arranged at n+1 equal divisions along the leaf from leaf root 1 to leaf tip 2, if the leaf length (i.e., the distance from leaf root 1 to leaf tip 2) is L, then the distance of each division is L / (n+1). Therefore, the distances r between each marker point 4 and the leaf root 1 are L / (n+1), 2L / (n+1), ..., nL / (n+1).

[0043] Specifically, each leaf has at least five marker points 4 to ensure accuracy during subsequent leaf morphology fitting. In this embodiment, each leaf has five marker points 4, as shown in the figure. Figure 1 The distances between the five marker points 4 and the leaf root 1 are r1 = L / 6, r2 = L / 3, r3 = L / 2, r4 = 2L / 3, and r5 = 5L / 6, respectively. Furthermore, the five marker points 4 are labeled with different colors, such as purple, blue, green, orange, and red, to facilitate image recognition.

[0044] S2. Capture video of the blades passing through tower 6. Based on the video information, identify and calculate the minimum distance d between each marker point 4 and tower 6, i.e., the minimum distance d between each marker point 4 and tower 6 is d1, d2, ..., d6 respectively. n ;

[0045] The specific calculation method for the minimum distance d between each marker point 4 and the tower 6 is as follows:

[0046] S2.1 After the camera 3 is installed, calibrate the conversion ratio of pixel distance in the video image to object distance; specifically, during calibration, the pixel position and object position of the marker point 4 are used as calibration points. Since the object distance between two adjacent marker points 4 is known, and the pixel distance between two adjacent marker points 4 can be obtained through video information, the conversion ratio of pixel distance in the video image to object distance is calibrated.

[0047] S2.2. Based on the video information, identify the pixel position of the marker point 4 on the blade in each frame of the image when the blade passes through the tower 6, fit the curve to obtain the pixel position of each marker point 4 closest to the tower 6, and identify the pixel position of the tower 6 in the horizontal direction when each marker point 4 is closest to the tower 6, so as to obtain the minimum pixel distance between each marker point 4 and the tower 6.

[0048] In this embodiment, based on historical experience, the curve of the blade marker point 4 passing through the tower 6 in each frame image is fitted to a straight line, such as... Figure 2 As shown, a linear fitting method is used to fit the marker point 4, thereby obtaining the pixel position of each marker point 4 closest to the tower 6. However, depending on the on-site operating conditions of the wind turbine, other fitting methods may be used.

[0049] S2.3. Based on the conversion ratio between pixel distance and actual distance, convert the minimum pixel distance between each marker point 4 and the tower 6 into the minimum distance d between each marker point 4 and the tower 6.

[0050] In this embodiment, the minimum distances between the five marker points 4 and the tower 6 are d1, d2, d3, d4, and d5, respectively.

[0051] S3. Calculate the planar position of each marker point 4 based on the distance r between each marker point 4 and the leaf root 1 and the minimum distance d between the corresponding marker point 4 and the tower 6.

[0052] The specific calculation method for the planar position of each marker point 4 is as follows:

[0053] S3.1. Taking the position of blade root 1 as the origin 0, establish the x-axis coordinate along the horizontal direction from the nose to the tail of the fuselage, and establish the y-axis coordinate along the vertical direction, as follows: Figure 3 As shown;

[0054] S3.2. Based on the wind turbine structural parameters, the abscissa of the point near the edge of the blade in tower 6 can be obtained as t0. (See...) Figure 3 ;

[0055] S3.3, Assume the coordinates of the i-th marker point 4 are (x... i y i Let i be an integer from 1 to n, and let r be the distance between the i-th marker 4 and the leaf root 1. i If we ignore the large deformation of marker point 4, then x i 2 +y i 2 =r i 2 ;

[0056] S3.4, Given that the minimum distance between the i-th marker point 4 and the tower 6 is di At this time, the coordinates of the point 4 corresponding to the i-th marker point 6 in the horizontal direction, which is the edge of the tower 6 closest to the blade, are (t0, y). i If x i =t0-d i ;

[0057] In this embodiment, the coordinates of the horizontal tower 6 near the edge of the blade corresponding to the five marker points 4 are (t0, y1), (t0, y2), (t0, y3), (t0, y4), and (t0, y5), respectively. (See...) Figure 3 ;

[0058] S3.5, The coordinates of the i-th marker point 4 are calculated as follows: Thus, the coordinates of each marker point 4 are obtained;

[0059] S4. Based on the planar position of each marker point 4 and the blade model, fit the blade morphology curve 5 and calculate the load.

[0060] Specifically, the methods for fitting the blade morphology curve 5 include linear regression, polynomial regression, locally weighted scatter smoothing, logistic regression, and nonlinear regression. In this embodiment, based on the blade morphology and vibration modes, the polynomial regression method is used to fit the blade morphology curve 5. The specific steps are as follows:

[0061] S4.1 Taking the third-order polynomial regression method as an example, then y = w0 + w1x + w2x 2 +w3x 3 ;

[0062] S4.2 Given the position of leaf root 1 and the planar positions of n marker points 4, substitute the coordinates of these n+1 points into the above formula, and according to the method of minimizing the error of scattered points, find the optimal parameters w0, w1, w2, w3 to obtain the leaf morphology curve.

[0063] After obtaining the blade morphology curve, the blade morphology curve is compared with the blade morphology when it is not under stress to obtain the deformation at each point, and then the current morphological load of the blade is obtained.

[0064] S5. Calculate the position of leaf tip 2 based on the leaf morphology curve;

[0065] Specifically, the position of leaf tip 2 is calculated based on the distance L between leaf tip 2 and leaf root 1 and the fitted leaf morphology curve 5;

[0066] Assume the coordinates of leaf tip 2 are (x tip y tip ),but:

[0067] x tip 2 +ytip 2 =L 2 ;

[0068] y tip =w0+w1x tip +w2x tip 2 +w3x tip 3 ;

[0069] Given L, w0, w1, w2, and w3 from the preceding steps, the coordinates of the leaf tip 2 can be calculated by solving the equation.

[0070] S6. Calculate the clearance value D based on the position of blade tip 2 and the wind turbine structural parameters; based on the wind turbine structural parameters, the x-coordinate of the point on the edge of the tower 6 closest to the blade is known to be t0. Obtain the x-coordinate of blade tip 2 from S5. tip The calculated net clearance value D is x. tip -t0.

[0071] In this invention, multiple marker points 4 are arranged on the blade. Video images are used to identify the minimum distance between each marker point 4 and the tower 5 when the blade passes the tower 6, thus obtaining the planar position of each marker point 4 when the blade is closest to the tower 6. Based on the planar positions of each marker point 4, the position information of the blade root 1, and the wind turbine structural parameters, a blade morphology curve when the blade is closest to the tower 6 is obtained. By comparing the blade morphology curve with the blade's morphology in a non-stressed state, the blade deformation is obtained, leading to the current morphological load condition of the blade. Simultaneously, the blade tip position and clearance value are calculated using the blade morphology curve. The clearance value, blade morphology changes, and load conditions are fed back to the main control system, which determines whether an early warning is needed and activates the clearance protection strategy.

Claims

1. A video-based method for identifying blade morphology and clearance based on multiple blade markers, characterized in that... Includes the following steps: S1. Deploy cameras near the blades in the nacelle. Arrange markers at n+1 equal division points between the blade root and the blade tip. Arrange n markers on each blade, and all markers are located on the leeward side of the blade. Record the distance r between each marker and the blade root in sequence. S2. Capture video of the blades passing through the tower, and identify and calculate the minimum distance d between each marker point and the tower based on the video information. The specific calculation method is as follows: S2.1 After the camera is installed, calibrate the conversion ratio between pixel distance in the video image and actual object distance; S2.

2. Based on the video information, identify the pixel position of the marker point on the blade in each frame of the image when the blade passes the tower, fit the curve to obtain the pixel position of each marker point closest to the tower, and identify the pixel position of the tower in the horizontal direction when each marker point is closest to the tower, so as to obtain the minimum pixel distance between each marker point and the tower. S2.

3. Based on the conversion ratio between pixel distance and actual distance, convert the minimum pixel distance between each marker point and the tower into the minimum distance d between each marker point and the tower. S3. Based on the distance r between each marker point and the leaf root, and the minimum distance d between the corresponding marker point and the tower, calculate the planar position of each marker point. The specific calculation method is as follows: S3.

1. With the blade root position as the origin 0, establish the x-axis coordinate along the horizontal direction from the nose to the tail of the nacelle, and establish the y-axis coordinate along the vertical direction. S3.

2. Based on the wind turbine structural parameters, the abscissa of the point on the edge of the tower near the blade is t0. S3.3, Assume the coordinates of the i-th marker are (x... i y i (i) is an integer from 1 to n, and the distance between the i-th marker and the leaf root is r. i Then x i 2 +y i 2 =r i 2 ; S3.4, Given that the minimum distance between the i-th marker and the tower is d i Then x i =t0-d i ; S3.5, The coordinates of the i-th marker point are calculated as ( , This allows us to obtain the coordinates of each marker point. S4. Based on the planar positions of each marker point and the blade model, fit the blade morphology curve and calculate the load. Based on the blade morphology and vibration modes, use a polynomial regression method to fit the blade morphology curve. The specific steps are as follows: S4.1, Using the third-order polynomial regression method, then ; S4.2 Given the leaf root position and the planar positions of n marker points, calculate the optimal parameters w0, w1, w2, and w3 using the method of minimizing the error of scattered points, and obtain the leaf morphology curve. The specific calculation method for the load condition is as follows: After obtaining the blade shape curve, the blade shape curve is compared with the shape of the blade in the non-stressed state to obtain the deformation of each point, and then the current shape load condition of the blade is obtained. S5. Calculate the leaf tip position based on the leaf morphology curve; S6. Calculate the headroom value based on the blade tip position and fan structural parameters.

2. The video recognition method for blade morphology and clearance based on multiple blade markers according to claim 1, characterized in that: Each leaf has no fewer than five markings.

3. The video recognition method for blade morphology and clearance based on multiple blade markers according to claim 1, characterized in that: In S1, the distance between each marker point and the leaf root is calculated based on the leaf length.

4. The video recognition method for blade morphology and clearance based on multiple blade markers according to claim 1, characterized in that: In S5, the position of the leaf tip is calculated based on the distance between the leaf tip and the leaf root and the fitted leaf morphology curve.

5. The video recognition method for blade morphology and clearance based on multiple blade markers according to claim 1, characterized in that: The airspace clearance value and load status are fed back to the main controller, which then determines whether an early warning is needed and activates the airspace protection strategy.

Images