A floating type fan state monitoring method based on visual vibration measurement

Abstract

The application discloses a floating wind turbine state monitoring method based on visual vibration measurement, and comprises the following steps: determining the visual camera layout position and quantity according to the floating wind turbine monitoring requirement, shooting the operation state of the floating wind turbine, pre-processing the original video, and outputting the pre-processed video signal; solving the motion displacement of each pixel point of each frame of the output video signal, and obtaining the full-field motion information; and realizing the accurate calculation of the floating platform motion and the tower vibration through the joint solving of multiple points in space. The application greatly improves the spatial resolution of the sensor monitoring, can realize full-field measurement, reduces the number of sensor arrangements, and simplifies the arrangement process and reduces the cost, because the visual camera serves as a non-contact sensor and does not produce additional mass load effect. The application can realize the synchronous measurement of the surge, heave, pitch of the floating wind turbine and the tower vibration, and provides an effective measurement means for the health state evaluation, fault diagnosis and operation and maintenance of the floating wind turbine.

Description

Technical Field

[0001] This invention relates to the field of floating wind turbine condition monitoring technology, and in particular to a floating wind turbine condition monitoring method based on visual vibration measurement. Background Technology

[0002] During their service life, floating wind turbines will face various marine environmental loads over long periods of time, which will seriously affect the stability and safety of their operation. In order to detect potential safety hazards in a timely manner, it is essential to monitor their system status response.

[0003] Floating wind turbines mount wind turbine generators on floating platforms, which are then positioned relative to each other using a mooring system. The movement of the floating platform causes a significant dynamic response in the turbine, and the stability of this movement is closely related to the power generation efficiency of the turbine generator and the overall system safety. The tower, as a crucial supporting structure of the floating wind turbine, houses the large turbine generator and connects to the floating platform; its structural reliability directly impacts the stable operation of the turbine. Therefore, the reliability and stability of the floating platform and tower structure are fundamental to the safe operation of the floating wind turbine system and are key to its long-term stable operation. Floating wind turbines operate in harsh offshore environments, facing complex wind, wave, and current loads. During operation, they are constantly subjected to environmental erosion, material aging, fatigue effects, and load effects, leading to continuous and cumulative structural damage and increasing the likelihood of sudden accidents.

[0004] The system response of a floating wind turbine can be reflected by multiple indicators, such as tilt angle and acceleration. Traditional monitoring systems using sensors such as accelerometers and fiber optics often need to be installed on the object being measured. This not only consumes a lot of manpower, material resources, and financial resources, but also imposes a mass load on the object, thus affecting the structural dynamic characteristics. Non-contact sensors generally rely on some kind of electromagnetic radiation to transmit vibration information. Commonly used methods include vibration measurement based on microwave interferometers, vibration measurement based on millimeter-wave sensors, and laser measurement methods. However, these methods are limited by the measurement range of the measuring instruments, and have problems such as not being able to perform multi-scale and full-field measurements, making it difficult to effectively monitor the state of large structures such as floating wind turbines. Vision-based motion measurement methods commonly used include digital image correlation, optical flow, and phase-based video motion estimation methods. Digital image correlation and optical flow methods directly extract the motion signal of the object from the original image. They have poor resistance to environmental noise and poor extraction effect in areas without obvious texture. Often, auxiliary markers need to be attached to the target object to improve the accuracy of recognition.

[0005] Unlike digital image correlation and optical flow methods, which directly manipulate pixel values ​​of the original image, phase-based motion estimation methods utilize two-dimensional Gabor filters of different directions and scales to map each frame of video to a new spatial frequency domain, and extract vibration signals through phase difference between consecutive frames. Therefore, phase-based video motion estimation methods have strong noise resistance and can identify object motion without the need for additional markers. However, phase-based video motion estimation methods can only identify single-point x and y-direction motion, identifying single-degree-of-freedom motion of objects such as cantilever beams and bridges by reading the motion history of the point of interest. In contrast, the motion response of floating wind turbines has multi-degree-of-freedom coupling characteristics, making state monitoring impossible.

[0006] Therefore, those skilled in the art are dedicated to developing a floating wind turbine condition monitoring method based on visual vibration measurement and to further exploring phase-based video motion estimation methods. Summary of the Invention

[0007] In view of the above-mentioned defects of the prior art, the technical problem to be solved by the present invention is that: in the condition monitoring of floating wind turbines, sensors can only monitor local and fixed-point vibration signals. Therefore, for large engineering structures such as floating wind turbines, a large number of sensors often need to be installed on the floating wind turbine structure, which will bring additional mass load effect to the structure, and is expensive and time-consuming and labor-intensive to arrange; the condition monitoring of floating wind turbines cannot identify the multi-degree-of-freedom coupling characteristics of floating wind turbines.

[0008] To achieve the above objectives, the present invention provides a method for monitoring the condition of a floating wind turbine based on visual vibration measurement, comprising the following steps:

[0009] Step 1: Determine the location and number of vision cameras according to the monitoring requirements of the floating wind turbine, capture the operating status of the floating wind turbine, and perform pre-processing such as cropping and noise reduction on the original video to output the pre-processed video signal.

[0010] Step 2: Solve for the motion displacement of each pixel in each frame of the video signal output in Step 1 to obtain the overall motion information;

[0011] Step 3: Through multi-point joint solution in space, further accurate calculation of the motion of the floating platform and the vibration of the tower is achieved.

[0012] Furthermore, in step 1, determining the location and number of visual cameras based on the monitoring requirements of the floating wind turbine includes determining them based on the actual wave direction and the direction of the wind turbine blades.

[0013] Furthermore, when the visual camera is positioned directly in front of or behind the wind turbine blades, it can monitor the swaying, heaving, and rolling motions of the floating platform, as well as the left-right vibration of the tower. When the visual camera is positioned directly to the side of the wind turbine blades, it can monitor the swaying, heaving, and rolling motions of the floating platform, as well as the front-back vibration of the tower.

[0014] Furthermore, in step 2, the motion displacement of each pixel in each frame of the video signal output in step 1 is solved. The phase-based video motion estimation method adopts the complex direction controllable pyramid algorithm.

[0015] Furthermore, step 2 includes the following steps:

[0016] Step 2.1: Determine a set of parameters for the Gabor wavelet in the 2D Gabor filter based on the video signal output in Step 1. The set of parameters includes the wavelength, direction, phase shift, Gaussian bandwidth, and aspect ratio of the Gabor wavelet.

[0017] Step 2.2: Convolve the Gabor wavelet with the video image to obtain a complex image with amplitude and phase;

[0018] Step 2.3: Obtain the phase difference of each pixel by assuming phase invariance;

[0019] Step 2.4: Calculate and output the motion information of the entire field using the least squares optimization criterion.

[0020] Furthermore, in step 2.1, the Gabor wavelet is a Gaussian-modulated complex sine function:

[0021]

[0022] In the formula, (x, y) are pixel coordinates; x′=xcosθ+ysinθ, y′=-xsinθ+ycosθ; λ, θ, ψ, σ, γ are a set of parameters representing the wavelength, direction, phase shift, Gaussian bandwidth, and aspect ratio of the 2D Gabor wavelet, respectively; j is the imaginary number sign, j 2 =-1.

[0023] Further, step 2.2 specifically includes: at time t, the image brightness value is I(x,y,t). By performing convolution operations with 2D Gabor wavelets of different scales and directions, the image I(x,y,t) is decomposed into the spatial frequency domain to obtain a complex image with amplitude and phase.

[0024]

[0025] In the formula: Q is the output complex image, C θA is the Gabor wavelet function in a 2D Gabor filter; r,θ For the local magnitude of a complex image, For local phase, r represents the number of layers in the complex pyramid; different scales of the complex pyramid are obtained by image downsampling, that is, the image size of each layer is reduced by a factor of 0.5, and the direction θ is either 2 or 4 directions, where 2 directions refer to the horizontal and vertical directions, and 4 directions are θ = 0°, θ = 45°, θ = 90°, and θ = 135°.

[0026] Further, step 2.3 specifically includes: assuming that the phase of the phase-based image remains unchanged, that is, the phase at (x,y,t0) remains unchanged after a small displacement (u(x,y,t0), v(x,y,t0)), performing a Taylor expansion and ignoring higher-order terms, the following expression is obtained:

[0027]

[0028] In the formula: The phase difference at position (x,y) contains the local motion displacement information (u,v) of the object;

[0029] Step 2.4 specifically includes: establishing a least squares optimization criterion to achieve accurate estimation of the local motion (u,v):

[0030]

[0031] After simplification, the analytical solution of the velocity matrix is ​​expressed as:

[0032] V=(Z Τ WZ) -1 (Z Τ WY)

[0033] In the formula: V represents the motion displacement matrix (u,v,t0), and Y represents the vector. W indicates that the diagonal elements are... A diagonal matrix, Z represents a matrix

[0034] The motion displacement of each pixel in each frame of the video is calculated to obtain the overall motion information.

[0035] Furthermore, step 3 includes the following steps:

[0036] Step 3.1: Select the motion calculation results of n points on the floating wind turbine, and use the least squares method to solve for the swaying, heaving, and pitching motions of the floating platform;

[0037] Step 3.2: Calculate the tower vibration based on the floating platform motion and tower displacement information;

[0038] Step 3.3: Output the motion information of the floating wind turbine platform and the vibration information of the tower.

[0039] Furthermore, step 3 specifically includes:

[0040] Treating the floating platform as a rigid body and the tower as a flexible cantilever beam, and establishing the coordinate system of the floating wind turbine at the center of gravity (x0, y0), the displacement (u, v) of point (x, y) under the coupled motions of swaying, heaving, and rolling of the floating platform is expressed as:

[0041]

[0042] In the formula, s, h, and p represent the swaying, heaving, and pitching motions of the floating platform, respectively. The trigonometric functions in the above formula are approximated, where sinp = p and cosp = 1 - 0.5p. 2 And ignore higher-order terms; automatically select n points on the floating platform using a script, and solve the resulting n sets of equations using least squares, the expression is:

[0043]

[0044] In the formula, w1 and w2 are weighting coefficients;

[0045] The sway, heave, and pitch motions of the floating platform at each moment are obtained by joint solution.

[0046] The displacement of a point (x, y) on the tower in the x-direction is a superposition of the pitching and swaying motions of the floating platform and the forward and backward vibrations of the tower itself, and can be expressed as:

[0047] u t =s+bcosp+(x-x0)cosp+(y-y0)sinp

[0048] In the formula: u t denoted by , where represents the absolute total displacement of the tower in the x-direction, and b represents the tower's own vibration.

[0049] The vibration information of the floating wind turbine tower is thus calculated.

[0050] Compared with the prior art, the present invention has the following main advantages:

[0051] (1) By using a phase-based visual vibration measurement method and a high spatial resolution visual camera sensor, non-contact, long-distance, high spatial resolution monitoring of floating wind turbines can be achieved, which greatly improves the spatial resolution of sensor monitoring and enables full-field measurement. The number of sensors is greatly reduced. As a non-contact sensor, the visual camera will not generate additional mass load effect, which simplifies the deployment process and reduces costs.

[0052] (2) Further processing of the vibration signals identified by the vision-based vibration processing method, and the identification and decomposition of the motion characteristics of each component of the multi-degree-of-freedom coupled motion characteristics of the floating wind turbine by adopting the two-dimensional motion analysis method, can realize the synchronous measurement of the swaying, heave, swaying, pitching and rolling motion of the floating wind turbine as well as the tower vibration, providing an effective measurement means for the health status assessment, fault diagnosis and operation and maintenance of floating wind turbines.

[0053] The following will further explain the concept, specific structure, and technical effects of the present invention in conjunction with the accompanying drawings, so as to fully understand the purpose, features, and effects of the present invention. Attached Figure Description

[0054] Figure 1 This is a flowchart of a preferred embodiment of the floating wind turbine motion vibration monitoring method based on visual vibration measurement according to the present invention;

[0055] Figure 2 This is a schematic diagram showing the relative positions of the visual camera and the floating wind turbine according to a preferred embodiment of the present invention;

[0056] Figure 3 This is a schematic diagram of the simulation video monitoring results of a floating wind turbine according to a preferred embodiment of the present invention;

[0057] Figure 4 This is a schematic diagram of the monitoring results of a floating wind turbine pool model test in a preferred embodiment of the present invention.

[0058] Among them, 21-vision camera, 22-floating wind turbine, 221-floating platform, 222-wind turbine, 223-mooring cable, 23-wave direction. Detailed Implementation

[0059] The preferred embodiments of the present invention are described below with reference to the accompanying drawings to make the technical content clearer and easier to understand. The present invention can be embodied in many different forms, and the scope of protection of the present invention is not limited to the embodiments mentioned herein.

[0060] In the accompanying drawings, components with the same structure are indicated by the same numerical designation, and components with similar structures or functions are indicated by similar numerical designations. The dimensions and thicknesses of each component shown in the drawings are arbitrary, and the present invention does not limit the dimensions and thicknesses of each component. To make the illustrations clearer, the thickness of some components has been appropriately exaggerated in the drawings.

[0061] This embodiment provides a method for monitoring the motion vibration of a floating wind turbine based on visual vibration measurement. The steps include: video preprocessing, full-field motion information identification, and multi-point joint motion solution. The method flow is as follows: Figure 1 As shown.

[0062] Figure 2 This is a schematic diagram showing the relative positions of the visual camera and the floating wind turbine. The floating wind turbine 22 consists of a floating platform 221, a wind turbine 222, and several mooring cables 223. The floating platform 221 is anchored in place by the mooring cables 223. The tower of the wind turbine 222 is fixedly installed on the floating platform 221. When the wind force is large at sea and the wind and wave directions are the same, the blades of the wind turbine 222 face the wave direction 23.

[0063] Video preprocessing. Based on the monitoring requirements of the floating wind turbine 22, the deployment positions and number of vision cameras 21 are determined to film the floating wind turbine 22. The video is then cropped according to the movement amplitude of the floating wind turbine 22 in the video, saving computing resources and reducing computational load while ensuring that no effective information is lost. When the vision camera 22 is positioned directly in front of / behind the wind turbine 222, it can monitor the swaying, heaving, and rolling movements of the floating platform 221, as well as the left-right vibration of the tower. When the vision camera 21 is positioned directly to the side of the wind turbine 222, it can monitor the swaying, heaving, and rolling movements of the floating platform 221, as well as the front-back vibration of the tower.

[0064] Full-field motion information recognition. Single-point motion extraction is performed using a phase-based video motion estimation method from vision-based vibration measurement techniques to solve for the global motion of the video. The phase-based video motion estimation method employs a complex-direction-controllable pyramid algorithm, which decomposes the video image into a series of sub-bands with different scales and directions using Gabor wavelets in a two-dimensional Gabor filter. The Gabor wavelet is a complex sine function modulated by Gaussian.

[0065]

[0066] In the formula, (x, y) are pixel coordinates; x′=xcosθ+ysinθ, y′=-xsinθ+ycosθ; λ, θ, ψ, σ, γ are a set of parameters representing the wavelength, direction, phase shift, Gaussian bandwidth, and aspect ratio of the 2D Gabor wavelet, respectively; j is the imaginary number sign, j 2 =-1.

[0067] The image brightness value at time t is I(x,y,t). By convolving it with 2D Gabor wavelets of different scales and directions, the image I(x,y,t) is decomposed into the spatial frequency domain, resulting in a complex image with amplitude and phase.

[0068]

[0069] In the formula: Q is the output complex image, C θ A is the Gabor wavelet function in a 2D Gabor filter;r,θ For the local magnitude of a complex image, For local phase, r represents the number of layers in the complex pyramid. Different scales of the complex pyramid are generally obtained by image downsampling, that is, the image size of each layer is reduced by a factor of 0.5. The direction θ is generally taken as 2 directions, 4 directions, etc., where 2 directions generally refer to the horizontal and vertical directions (θ = 0° and θ = 90°), 4 directions are θ = 0°, θ = 45°, θ = 90°, θ = 135°, and so on.

[0070] Assuming the phase of the phase-based image remains unchanged, i.e., the phase at (x,y,t0) remains unchanged after a small displacement (u(x,y,t0), v(x,y,t0)), a Taylor expansion is performed, ignoring higher-order terms, yielding the following expression:

[0071]

[0072] In the formula: The phase difference at position (x,y) contains the local motion displacement information (u,v) of the object.

[0073] To achieve accurate estimation of local motion (u,v), a least squares optimization criterion is established:

[0074]

[0075] After simplification, the analytical solution of the velocity matrix can be expressed as:

[0076] V=(Z Τ WZ) -1 (Z Τ WY)

[0077] In the formula: V represents the motion displacement matrix (u,v,t0), and Y represents the vector. W indicates that the diagonal elements are... A diagonal matrix, Z represents a matrix Based on the above formula, the motion displacement of each frame and each pixel in the video can be solved, thereby obtaining the motion information of the entire field.

[0078] Multi-point joint solution of motion. Phase-based video motion estimation methods can obtain the 2-DOF motion displacement (u,v) at each pixel (x,y). The motion response of floating wind turbines has the characteristics of multi-DOF coupling, which requires joint solution of multiple points in space to further achieve accurate solution of the motion of the floating platform 221 and the vibration of the tower.

[0079] Treating the floating platform 221 as a rigid body and the tower as a flexible cantilever beam, and establishing the coordinate system of the floating wind turbine at the center of gravity (x0, y0), the displacement (u, v) of point (x, y) of the floating platform 221 under the coupled motions of swaying, heaving, and rolling can be expressed as:

[0080]

[0081] In the formula, s, h, and p represent the sway, heave, and pitch motions of the floating platform 221, respectively. Since the pitch motion of the floating fan is relatively small, it can be approximated. The trigonometric functions in the above formula are approximated, where sinp = p, cosp = 1 - 0.5p. 2 And higher-order terms are ignored. Meanwhile, considering the instability of single-point calculation results, the fact that the above equation is an underdetermined system of equations, and the limited computational resources, a script automatically and randomly selects n points on the floating platform 221 to solve the resulting n sets of equations using least-squares, resulting in the expression:

[0082]

[0083] In the formula, w1 and w2 are weighting coefficients. For example, we can assume that w1 = w2 = 0.5. The sway, heave, and pitch motions (s, h, p) of the floating platform 221 at each moment can be obtained by combining these equations.

[0084] Since the floating wind turbine tower is mounted on the floating platform 221, the tower structure will vibrate due to the movement of the floating platform, the vibration of the nacelle and rotor, and environmental loads. The displacement of a point (x, y) on the tower in the x-direction is a superposition of the pitching and swaying motions of the floating platform 221 and the forward and backward vibrations of the tower itself, and can be expressed as:

[0085] u t =s+bcosp+(x-x0)cosp+(y-y0)sinp

[0086] In the formula: u t Let represent the absolute total displacement of the tower in the x-direction, and b represent the tower's own vibration. At this point, all parameters except for the tower's own vibration b have been calculated. This formula provides the vibration information of the floating wind turbine tower.

[0087] Figure 3This is a schematic diagram of the simulation video monitoring results for a floating wind turbine. The simulation video simulates the swaying, heaving, and pitching motions of the floating platform 221, as well as the forward and backward vibrations of the tower. It simulates a 1:50 equivalent scale model of a floating wind turbine. The relevant characteristic parameters and motion / vibration characteristics correspond to those of a real floating wind turbine, and the tower is set with first and second order vibration frequencies. Its motion trajectory is set according to the actual measured operating conditions of the target floating wind turbine under white noise excitation. As can be seen from the time-series diagram, the identification results of this embodiment highly overlap with the reference values, with the error remaining within 1%. From the frequency domain response, it can be seen that the motion response of the floating wind turbine under the action of the white noise wave is mainly the wave frequency, f. t1 f t2 f represents the first and second order vibration frequencies of the tower, respectively. w This represents the frequency of the waves. For all identified tower top displacements, in addition to the motion frequencies contained in the floating platform 221, the first and second order vibration frequencies of the wind turbine tower itself can also be accurately identified.

[0088] In actual floating wind turbine status monitoring, complex environmental disturbances such as wave undulations can affect the images captured by the visual camera, thus impacting the accuracy of motion / vibration calculations. To evaluate the robustness and anti-interference capability of the method in this embodiment, a simulated video was used as a prototype, and digital Gaussian noise of different intensities was added to the video to simulate environmental noise. The mean of Gaussian noise affects the brightness of the video, while the standard deviation directly affects the clarity and detail of the video. This embodiment only discusses the impact of changes in video clarity, i.e., changes in the noise standard deviation, on recognition accuracy. Therefore, in the four noise intensity levels, the mean of Gaussian noise was set to 0, and the noise intensity was represented by the standard deviation α, which were 0.01, 0.05, 0.1, and 0.2, respectively. The results show that the recognition error of the system motion response increases with the increase of Gaussian noise. The recognition errors of the swaying, heaving, and tower top displacement motions of the floating platform 221 are less affected by noise compared to the calculation results from the noise-free simulation video; when the standard deviation of noise interference is below 0.2, the recognition error remains within 5%. However, noise has a significant impact on the recognition results of the swaying motion of the floating platform 221, increasing the error to 3.6%-38%. This is because Gaussian noise reduces the accuracy of single-point solutions, and since the swaying motion is a small value in the solution formula, fluctuations in the single-point motion recognition results have a significant impact on the results. However, it has a smaller impact on larger motions such as the floating wind turbine platform motion and tower displacement. These results indicate that even under strong background noise interference, the method of this embodiment can still accurately identify large-scale key dynamic responses such as the floating wind turbine platform motion and tower displacement, but the accuracy for smaller-scale motion recognition is somewhat lacking. The method of this embodiment shows good robustness in the recognition of large-scale motions in videos with a certain intensity of Gaussian noise.

[0089] Figure 4This is a schematic diagram of the monitoring results of a floating wind turbine model test in a water tank. The analysis object is a scaled-down test model of a floating wind turbine in a wind, wave, and current test tank. A non-contact optical measurement system was set up on the floating platform to capture the six degrees of freedom motion of the floating platform and use it as a reference for the identification results. However, there were no high-precision tower vibration measurement results for reference in the experiment. White noise wave excitation was selected for the identification and analysis of the floating wind turbine system response. The wave height was 2.45 cm, and the wave period was 0.625-3.75 s. A camera was installed on one side of the water tank to record the response of the floating wind turbine model at a frame rate of 25. To save computing resources, the video resolution was cropped to 340×620 for processing, with one pixel representing an actual distance of 4.121 mm. It can be seen from the time-history graph that the motion trajectory recognition is good, but the matching of motion amplitude is slightly worse, with an average error within 10%. This is partly because non-contact optical measurement cannot accurately identify the motion of the floating platform. The results show that, even in complex and highly disturbed floating wind turbine operating scenarios, the phase-based video motion estimation method proposed in this invention still maintains high recognition accuracy. It achieves long-distance, non-contact, full-field, multi-scale, and unstructured load-accurate monitoring of floating wind turbines, a capability unattainable by traditional monitoring methods.

[0090] The preferred embodiments of the present invention have been described in detail above. It should be understood that those skilled in the art can make numerous modifications and variations based on the concept of the present invention without creative effort. Therefore, all technical solutions that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of existing technology should be within the scope of protection defined by the claims.

Claims

1. A method for monitoring the condition of a floating wind turbine based on visual vibration measurement, characterized in that, Includes the following steps: Step 1: Determine the location and number of vision cameras according to the monitoring requirements of the floating wind turbine, capture the operating status of the floating wind turbine, and perform pre-processing such as cropping and noise reduction on the original video to output the pre-processed video signal. Step 2: Solve for the motion displacement of each pixel in each frame of the video signal output in Step 1 to obtain the overall motion information; Step 3: Through joint solution at multiple points in space, further accurate calculations of the floating platform motion and tower vibration are achieved; in, In step 2, the motion displacement of each pixel in each frame of the video signal output in step 1 is solved. The phase-based video motion estimation method uses the complex direction controllable pyramid algorithm. Step 2 includes the following steps: Step 2.1: Determine a set of parameters for the Gabor wavelet in the 2D Gabor filter based on the video signal output in Step 1. The set of parameters includes the wavelength, direction, phase shift, Gaussian bandwidth, and aspect ratio of the Gabor wavelet. Step 2.2: Convolve the Gabor wavelet with the video image to obtain a complex image with amplitude and phase; Step 2.3: Obtain the phase difference of each pixel by assuming phase invariance; Step 2.4: Calculate and output the full-field motion information using the least squares optimization criterion; In step 2.1, the Gabor wavelet is a Gaussian-modulated complex sine function: In the formula, These are pixel coordinates; , ; These are a set of parameters representing the wavelength, direction, phase shift, Gaussian bandwidth, and aspect ratio of a 2D Gabor wavelet. It is the symbol for imaginary numbers. ; Step 2.2 specifically includes: in Image brightness value at time By performing convolution operations with 2D Gabor wavelets of different scales and directions, the image is... Decomposing to the spatial frequency domain yields a complex image with amplitude and phase: In the formula: For the output complex image, For the Gabor wavelet function in a 2D Gabor filter; For the local magnitude of a complex image, For local phase, This indicates the number of layers in the complex pyramid; different scales of the complex pyramid are obtained by image downsampling, i.e., the image size of each layer is reduced by a factor of 0.5, and the orientation... Choose 2 or 4 directions, where 2 directions refer to the horizontal and vertical directions, and 4 directions are... ; Step 2.3 specifically includes: assuming that the phase of the phase-based image remains unchanged, that is... The phase at that point undergoes a tiny displacement After the phase remains unchanged, performing a Taylor expansion and ignoring higher-order terms yields the following expression: In the formula: For position The phase difference at that point contains information about the object's local motion and displacement. ; Step 2.4 specifically includes: to achieve local motion For accurate estimation, establish the least squares optimization criterion: After simplification, the analytical solution of the velocity matrix is ​​expressed as: In the formula: Represents the motion displacement matrix , Representing vectors , Indicates that the diagonal elements are diagonal matrix, Representation matrix ; The motion displacement of each pixel in each frame of the video is calculated to obtain the overall motion information. Step 3 includes the following steps: Step 3.1: Select the floating fan. n The point motion calculation results are used to solve the swaying, heaving, and pitching motions of the floating platform using least squares. Step 3.2: Calculate the tower vibration based on the floating platform motion and tower displacement information; Step 3.3: Output the motion information of the floating wind turbine platform and the vibration information of the tower. Step 3 specifically includes: The floating platform is treated as a rigid body, while the tower is treated as a flexible cantilever beam, and the coordinate system of the floating wind turbine is established at the center of gravity. At that point, the floating platform, under the coupled motion of swaying, heaving, and rolling, ... displacement Expressed as: In the formula, , , Let represent the swaying, heaving, and pitching motions of the floating platform, respectively; approximate the trigonometric functions in the above equation, where . And ignore higher-order minor terms; automatically and randomly select from the floating platform via script. A point, for the obtained Solving the system of equations using least squares, the expression is: In the formula, , These are the weighting coefficients; The combined solution yields the sway, heave, and pitch motions of the floating platform at each moment. ; For a certain point on the tower The displacement in the x-direction is a superposition of the pitching and swaying motions of the floating platform and the forward and backward vibrations of the tower itself, and can be expressed as: In the formula: This represents the absolute total displacement of the tower in the x-direction. This indicates the vibration of the tower itself; The vibration information of the floating wind turbine tower is thus calculated.

2. The floating wind turbine condition monitoring method based on visual vibration measurement as described in claim 1, characterized in that, In step 1, determining the location and number of visual cameras based on the monitoring requirements of the floating wind turbine includes determining them based on the actual wave direction and the direction of the wind turbine blades.

3. The floating wind turbine condition monitoring method based on visual vibration measurement as described in claim 2, characterized in that, When the vision camera is positioned directly in front of or behind the wind turbine blades, it can monitor the swaying, heaving, and rolling motions of the floating platform, as well as the left-right vibration of the tower. When the vision camera is positioned to the side of the wind turbine blades, it can monitor the longitudinal swaying, heaving, and rolling motions of the floating platform, as well as the front-back vibration of the tower.

Images