Method and System for Identifying the Eddy-Induced Vibration Response of Flexible Members in Transmission Towers

Abstract

A method and system for identifying vortex-induced vibration response of flexible transmission tower members is disclosed. The method includes: acquiring a vibration video stream of the flexible transmission tower member; decomposing the video stream into a frame sequence; extracting the region of interest (ROI) at the edge of the member; using bicubic interpolation to magnify each frame to obtain sub-pixel-level coordinates of the member edges; retaining edges that satisfy dual threshold constraints to obtain vibration displacement time-history data; extracting member appearance parameters and structural vibration parameters; performing scale calibration using the time baseline parallax method and multi-source data fusion method; calculating vibration intensity based on envelope analysis and moving root mean square (RMS) to identify the initiation and duration of vortex-induced vibration; using STFT to perform time-frequency analysis on the displacement signal to determine the frequency-locking characteristics of vortex-induced vibration; constructing a two-dimensional single-degree-of-freedom fluid-structure interaction numerical model; solving the structural motion equations; simulating the amplitude, frequency, and wake vortex-induced evolution under different reduced wind speeds; and verifying the identification results. This invention achieves high-precision, interference-resistant, fully automated, and mechanistically verifiable vibration monitoring.

Description

Technical Field

[0001] This invention belongs to the field of image and video pattern recognition and structural health monitoring, and specifically relates to a method and system for identifying the vortex-induced vibration response of flexible members of transmission towers based on machine vision. Background Technology

[0002] With the rapid development of ultra-high voltage power grids, steel pipe tower structures are widely used in transmission towers. The flexible members (tower leg diagonal members, horizontal members, and diaphragm members) have a large slenderness ratio, making them highly susceptible to vortex-induced vibration under light wind conditions. This vibration is characterized by low amplitude and high cycle number. Long-term effects can lead to cracking of gusset plates, loosening of bolts, fatigue failure of members, and even tower collapse.

[0003] Traditional monitoring solutions employ contact sensors such as accelerometers and fiber Bragg gratings, which suffer from high installation risks, high costs, difficult maintenance, and susceptibility to electromagnetic interference, failing to meet the needs of large-scale, long-term online monitoring. While machine vision can achieve non-contact monitoring, traditional methods have the following shortcomings: 1. Conventional edge detection can only achieve integer pixel-level positioning, while the amplitude of vortex-induced vibration in actual monitoring is usually at the millimeter level, corresponding to sub-pixel or even one-tenth of a pixel level. Traditional methods cannot effectively capture micro-amplitude vibration signals due to excessive quantization errors, resulting in insufficient sensitivity and inability to identify millimeter-level micro-amplitude vibrations. 2. Severe false edge interference in complex backgrounds. In actual engineering, changes in lighting, water reflection, background textures, etc., can easily generate a large number of false edges. Traditional template matching and optical flow methods are prone to target loss or mistracking under such interference. Moreover, pixel-level errors accumulate over time, causing significant drift in displacement time history and affecting the accuracy of spectrum analysis. 3. Lack of automatic identification of vortex-induced vibration initiation and quantitative analysis of frequency locking characteristics; Vortex-induced vibration has clear initiation wind speed threshold and frequency locking range characteristics, but existing methods mostly rely on manual experience to set parameters, lacking systematic analysis means for automatic initiation identification, quantitative identification of frequency locking platforms and multimodal vibration separation, which makes it difficult to support the needs of intelligent early warning.

[0004] 4. The mechanism was not verified by CFD numerical simulation. Most of the data was purely data-driven and CFD numerical simulation was not introduced into the verification closed loop. As a result, the extracted vibration frequency and amplitude lacked the support of fluid-structure interaction mechanism, making it difficult to confirm that the identified signal is indeed a vortex-induced response. The engineering credibility of the identification results was insufficient. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a machine vision-based method and system for identifying the vortex-induced vibration response of flexible members in transmission towers, thereby solving the technical problem of achieving high-precision, interference-resistant, fully automated, and mechanism-verifiable vibration monitoring.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution.

[0007] This invention first discloses a machine vision-based method for identifying the vortex-induced vibration response of flexible members in transmission towers. The method includes the following steps: Step 1: Use industrial image acquisition equipment to acquire the vibration video stream of the flexible pole of the transmission tower, decompose the video stream into an ordered frame sequence, extract the region of interest at the edge of the pole from the frame sequence, use bicubic interpolation to magnify frame by frame, and obtain the subpixel level coordinates of the pole edge based on the subpixel edge localization method; Step 2: Combining the double threshold constraint edge tracking algorithm, retain the edges that simultaneously satisfy the inter-frame consistency constraint and the spatial continuity constraint as real edges and remove false edges to obtain continuous and stable vibration displacement time history data. Based on the sub-pixel level coordinates, extract the appearance parameters and structural vibration parameters of the rod. Step 3: Based on the appearance parameters and structural vibration parameters, scale calibration is performed using the time baseline parallax method and multi-source data fusion method to establish the conversion relationship between pixel displacement and actual displacement. Vibration intensity is calculated based on Hilbert envelope analysis and sliding root mean square (RMS) analysis. An adaptive threshold discrimination mechanism is constructed with the vibration-free section as the baseline to automatically identify the vortex-induced vibration initiation and continuous sections. Step 4: Use Short Time Fourier Transform (STFT) to perform time-frequency analysis on the displacement signal, and determine the frequency locking characteristics of vortex-induced vibration based on vibration energy and dominant frequency; Step 5: Construct a two-dimensional CFD single-degree-of-freedom fluid-structure interaction numerical model based on the structural vibration parameters of the members. Use the SST k-ω turbulence model, dynamic mesh technology and fourth-order Runge-Kutta method to solve the structural motion equations. Simulate the amplitude, frequency and wake vortex-induced evolution under different reduced wind speeds to verify the identification results and vortex-induced vibration mechanism. Assess the risk of vortex-induced vibration and output operation and maintenance recommendations based on the assessment level.

[0008] The present invention further includes the following preferred embodiments: In step 1, the step of extracting the region of interest (ROI) of the rod edge from the frame sequence and performing frame-by-frame magnification using bicubic interpolation further includes: Select a rectangular region of interest (ROI) that includes the boundaries of the poles, avoiding interference areas caused by conductors, insulators, and tower intersecting components; apply bicubic interpolation to the ROI region to suppress jagged edges and blurring, while maintaining the continuity of edge grayscale gradients.

[0009] In the formula, w(i,j) is the bicubic interpolation weight function, I(x,y) is the original ROI image, and I'(x',y') is the magnified image.

[0010] The subpixel-level coordinates of the rod edge obtained by the subpixel edge localization method further include: Construct a 5×3 local window centered on the target pixel, and calculate the sum of gray levels S in the left, middle, and right columns respectively. L S M S R Sub-pixel localization is achieved by inferring edge line parameters based on the pixel area segmentation model.

[0011]

[0012] Where A and B are the grayscale values ​​of the rod and the background, h is the pixel side length, and E is the grayscale value of the background. L E M E R Let a and b be the area enclosed by the edge line within the corresponding column, where a and b are the parameters of the edge line.

[0013] Step 2 further includes: Combining a dual-threshold edge tracking algorithm to eliminate false edge interference, calculate the inter-frame consistency constraint D1 and the spatial continuity constraint D2:

[0014] y c (i) represents the y-coordinate of the edge point in the current frame. p The y-coordinate of the edge point corresponding to the previous frame. c (i+1) is the ordinate of the corresponding edge point in the next frame; Set thresholds T1=0.05 pixels and T2=0.2 pixels, retain only edge points that satisfy D1≤T1 and D2≤T2 as true edges, iterate frame by frame to track, remove false edges caused by noise, background structure and sudden changes in lighting, and output a continuous and smooth displacement time history.

[0015] The step of scaling by using the time baseline parallax method and multi-source data fusion method to establish the conversion relationship between pixel displacement and actual displacement further includes: A mapping relationship between image coordinates and actual physical coordinates is established. The time baseline parallax method is used for pixel and actual displacement scale calibration. The physical length L of the rod is measured, and the image pixel length p is read. L Calculate the scaling factor λ = L / p L To correct the tilt error between the physical plane and the imaging plane, a multi-source data fusion method is employed to fuse the displacement from the accelerometer sensor. The conversion coefficients are then optimized using the least squares method to construct the objective function.

[0016] The optimized conversion coefficients are obtained by solving:

[0017] In the formula, J(α) is the objective function, α is the transformation coefficient, N is the number of data samples, and u i For the accelerometer to measure displacement values, x i Extract displacement values ​​from the image.

[0018] The vibration intensity calculation based on Hilbert envelope analysis and sliding root mean square (RMS) further includes an adaptive threshold discrimination mechanism that uses the vibration-free section as a baseline to automatically identify the initiation and continuation sections of vortex-induced vibration. Perform a Fast Fourier Transform (FFT) on the calibrated displacement-time history to extract the dominant frequency f0 of the vortex-induced vibration and plot the power spectrum.

[0019] Narrowband bandpass filtering with a value of 0.7f0 to 1.3f0 is used to filter out low-frequency drift and high-frequency noise; Performing a Hilbert transform on the filtered signal yields the analytic signal and the instantaneous envelope:

[0020] The sliding RMS is calculated using a window of 5 to 10 times the vibration period; the sliding root mean square RMS is calculated as follows:

[0021] N is the number of data points within the sliding window, Δt is the sampling interval, and x(t-kΔt) is the vibration displacement value at the corresponding time. Using data from vibration-free sections as the baseline, the baseline mean μ and standard deviation α are calculated to construct an adaptive vibration initiation threshold:

[0022] In the formula, k is taken as 4; when the sliding RMS continuously exceeds the above threshold and remains for a preset time, it is determined that vortex-induced vibration has entered.

[0023] Step 4 further includes: The vibration displacement signal is subjected to a short-time Fourier transform (STFT) for time-frequency analysis and frequency locking determination.

[0024] x(τ) is the vibration displacement signal, w(τ-t) is the window function, f is the frequency, and j is the imaginary unit. If the vibration energy is concentrated in a single frequency band and the main frequency fluctuation is <0.3Hz, it is determined to be a frequency-locked state.

[0025] Step 5 further includes: Based on the extracted vibration parameters of the flexible rod structure, and combining fluid-structure interaction theory and structural dynamics principles, a two-dimensional single-degree-of-freedom fluid-structure interaction numerical model of the transmission tower's flexible rod is established. The three-dimensional spatial structure of the transmission cylindrical flexible steel pipe rod is equivalent to a two-dimensional crosswind single-degree-of-freedom vibration system, ignoring along-wind vibration and higher-order bending modes, retaining only the first-order dominant mode shape. Based on Euler-Bernoulli beam theory and the modal superposition principle, an amplitude amplification factor is introduced to correct the modal effect, and the structural dynamics equations are established.

[0026] m is the equivalent mass per unit length of the rod, c is the equivalent damping coefficient, and k is the equivalent stiffness coefficient. These represent the crosswind acceleration, velocity, and displacement of the member, respectively, F. L (t) represents the fluid force; This is the amplitude amplification factor for the first mode shape; The fluid domain uses SST The turbulence model uses a computational domain with a transverse width of 25D and a wake length of 32D, where D is the outer diameter of the rod. A structured mesh is used throughout the domain, with localized mesh refinement in the near-wall region of the cylinder. The first mesh layer near the wall satisfies the dimensionless wall distance. To capture the characteristics of boundary layer separation and Karman vortex-induced shedding; among which This is a general dimensionless wall distance in CFD, representing the normalized normal distance from the first-layer mesh node near the wall to the bar wall.

[0027] The structural motion equations are solved using a dynamic mesh, a user-defined function (UDF), and the fourth-order Runge-Kutta method.

[0028] In the formula: This refers to the time step for numerical computation. This is the acceleration slope term; This is the velocity slope term; The velocity and displacement at the current moment; Update velocity and displacement for the next moment; Calculate the reduced wind speed and dimensionless amplitude:

[0029] U represents the actual wind speed, f n Let A be the natural frequency of the rod, D be the diameter of the rod, and A be the diameter of the rod. max For the maximum vibration amplitude, U r To reduce wind speed, A The dimensionless amplitude is the equivalent mass per unit length of the rod.

[0030] Step 5 further includes: Output the start and end times of eddy-induced vibration, vibration amplitude, dominant frequency, frequency locking range, and hazard level; and provide graded operation and maintenance suggestions based on fatigue characteristics, including close monitoring, restricting the passage of large vehicles, installing turbulence devices, local maintenance and reinforcement, or line shutdown.

[0031] This invention also discloses a machine vision-based system for identifying the vortex-induced vibration response of flexible transmission tower members, utilizing the aforementioned machine vision-based method for identifying the vortex-induced vibration response of flexible transmission tower members, comprising: The image preprocessing module is used to acquire the vibration video stream of the flexible pole of the transmission tower using industrial image acquisition equipment, decompose the video stream into an ordered frame sequence, extract the region of interest at the edge of the pole from the frame sequence, perform frame-by-frame magnification using bicubic interpolation, and obtain the subpixel level coordinates of the pole edge based on the subpixel edge localization method. The edge parameter extraction module is used to combine the double threshold constraint edge tracking algorithm to retain the edges that simultaneously satisfy the inter-frame consistency constraint and the spatial continuity constraint as real edges and remove false edges to obtain continuous and stable vibration displacement time history data. Based on the sub-pixel level coordinates, the appearance parameters and structural vibration parameters of the rod are extracted. The vortex-induced vibration identification module is used to perform scale calibration based on the appearance parameters and structural vibration parameters by using the time baseline parallax method and multi-source data fusion method, establish the conversion relationship between pixel displacement and actual displacement, calculate vibration intensity based on Hilbert envelope analysis and sliding root mean square (RMS), and construct an adaptive threshold discrimination mechanism with the vibration-free section as the baseline to automatically identify the vortex-induced vibration initiation and continuous sections. The frequency locking characteristic determination module is used to perform time-frequency analysis on the displacement signal using short-time Fourier transform (STFT) and determine the frequency locking characteristic of vortex-induced vibration based on vibration energy and dominant frequency. The identification result verification module is used to construct a two-dimensional CFD single-degree-of-freedom fluid-structure interaction numerical model based on the structural vibration parameters of the members. It uses the SST k-ω turbulence model, dynamic mesh technology and the fourth-order Runge-Kutta method to solve the structural motion equations, simulate the amplitude, frequency and wake vortex-induced evolution under different reduced wind speeds, verify the identification results and vortex-induced vibration mechanism, assess the risk of vortex-induced vibration, and output operation and maintenance suggestions based on the assessment level.

[0032] Accordingly, this application also discloses a terminal, including a processor and a storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the aforementioned machine vision-based method for identifying the vortex-induced vibration response of flexible transmission tower members.

[0033] Accordingly, this application also discloses a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the aforementioned machine vision-based method for identifying the vortex-induced vibration response of flexible members of transmission towers.

[0034] The beneficial effects of this invention are as follows: Compared with existing technologies, this invention provides a machine vision-based method and system for identifying the vortex-induced vibration response of flexible transmission tower members. It eliminates the need for deploying numerous sensors and lines, significantly reducing installation and maintenance costs. Furthermore, it improves measurement accuracy through sub-pixel positioning technology, effectively avoiding subjective errors from manual observation and achieving automated and continuous monitoring. This method can not only identify structural parameters such as displacement and frequency of vortex-induced vibration in the members, but also perceive surface corrosion, fatigue, and other defects through image colorimetric analysis, achieving simultaneous monitoring of appearance damage and vibration damage. Simultaneously, it requires no modification to the tower structure; vibration data can be acquired over long distances and at multiple locations using only industrial image acquisition equipment, adapting to complex field environments. Employing adaptive thresholding and wavelet denoising techniques, parameters can be automatically adjusted to cope with changes in lighting and environmental noise, improving the accuracy and robustness of identification under complex operating conditions. Furthermore, this invention constructs a multi-index-based vortex-induced vibration risk assessment system, comprehensively judging vibration based on indicators such as vibration amplitude, dominant frequency, and frequency-locked interval, supporting tiered operation and maintenance and traffic control. By combining time-frequency analysis and CFD numerical simulation, vibration data is integrated with fluid-structure interaction mechanism verification, providing reliable data support for operation and maintenance decisions. This method can also be extended to the synchronous monitoring of vibration and defects in tall structures such as wind turbine towers and communication towers, as well as various flexible components, providing strong technical support for intelligent operation and maintenance of power infrastructure. Attached Figure Description

[0035] Figure 1 This is a flowchart of the machine vision-based method for identifying the vortex-induced vibration response of flexible transmission tower members in this invention.

[0036] Figure 2 This is a schematic diagram of video acquisition and ROI region selection for flexible transmission tower members in this invention; Figure 3 This is a schematic diagram illustrating the sub-pixel edge positioning principle based on area effect in this invention. Figure 4 This is a schematic diagram of the dual-threshold constraint edge tracking algorithm in this invention. Detailed Implementation

[0037] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0038] The embodiments described in this application are merely some, not all, embodiments of the present invention. Based on the spirit of the present invention, other embodiments obtained by those skilled in the art without inventive effort are all within the protection scope of the present invention.

[0039] To address the problems of existing technologies for monitoring vortex-induced vibration of flexible transmission tower members, which rely on traditional sensors, have high deployment costs, poor adaptability to complex environments, insufficient identification accuracy, and lack a full-process early warning and maintenance mechanism, this invention proposes a machine vision-based method and system for identifying the vortex-induced vibration response of flexible transmission tower members. By combining machine vision technologies such as non-contact measurement, sub-pixel positioning, time-frequency analysis, and time-series prediction, this invention achieves efficient and accurate identification, risk assessment, and hierarchical operation and maintenance management of vortex-induced vibration of flexible transmission tower members.

[0040] See Figure 1 As shown, the machine vision-based method for identifying the vortex-induced vibration response of flexible transmission tower members disclosed in this invention includes the following steps: Step 1: Use industrial image acquisition equipment to acquire the vibration video stream of the flexible pole of the transmission tower, decompose the video stream into an ordered frame sequence, extract the region of interest at the edge of the pole from the frame sequence, use bicubic interpolation to magnify frame by frame, and obtain the subpixel level coordinates of the pole edge based on the subpixel edge localization method.

[0041] Specifically, the data collection for transmission tower monitoring is divided into video data of the tower's appearance and data sets of structural vibration, such as... Figure 2 As shown, the vibration information of the rod edge is stored in a video file as an image sequence. Subsequently, the sub-pixel edge coordinates, vibration displacement time history, vibration dominant frequency, frequency locking interval, and vortex-induced vibration initiation characteristics of the rod will be extracted from the file.

[0042] During the shooting process, the camera's focal length, aperture, shutter speed, and white balance parameters were optimized. The resolution was set to 1920×1080 and the frame rate to 100fps. The target flexible pole was placed in the center of the frame, avoiding interference areas caused by conductors, insulators, and tower intersecting components. The region of interest (ROI) was selected. At the end of the acquisition phase, the camera was turned off and invalid segments with initial and final shaking, overexposure, and underexposure were removed. Data from stable vibration segments were retained to ensure the accuracy of subsequent parameter extraction.

[0043] In the monitoring area of ​​the flexible members of the transmission tower, a stable tracking section is determined, and the stationary nodes, bolt holes and fixed components of the tower body are selected as reference points. These are used as the reference coordinate origin for image displacement correction and anti-shake processing, as shown in Figure 3.

[0044] The acquired video images are processed by frame segmentation. The video frame segmentation algorithm in OpenCV is called, and two parameters are passed in: the video path to be segmented and the file path to save the images. The image is read frame by frame from the video file and an ordered frame sequence is generated.

[0045] Bicubic interpolation is performed on the ROI region to suppress jagged edges and maintain the continuity of the edge grayscale gradient. The interpolation formula is as follows:

[0046] w(i,j) represents the bicubic interpolation weights, I(x,y) represents the original ROI image, and I'(x',y') represents the enlarged image.

[0047] Next, based on the gray-level gradient difference, a sub-pixel edge localization algorithm based on area effect is used to extract the precise coordinates of the flexible rod edge, including the precise position of the rod edge, and to cut off the non-interest areas, thus completing the edge feature extraction. A 5×3 local window is constructed with the target pixel as the center, and the sum of gray levels S in the left, middle, and right columns is calculated respectively. L S M S R Sub-pixel localization is achieved by inversely calculating edge line parameters based on the pixel area segmentation model, as shown in the following formula:

[0048]

[0049] Where A and B are the grayscale values ​​of the rod and the background, h is the pixel side length, and E is the grayscale value of the background. L E M E R Let a and b be the area enclosed by the edge line within the corresponding column, where a and b are the parameters of the edge line.

[0050] Step 2: Combining the double threshold constraint edge tracking algorithm, retain the edges that simultaneously satisfy the inter-frame consistency constraint and the spatial continuity constraint as real edges and remove false edges to obtain continuous and stable vibration displacement time history data. Based on the sub-pixel level coordinates, extract the appearance parameters and structural vibration parameters of the rod.

[0051] Combining a dual-threshold edge tracking algorithm to eliminate false edge interference, calculate the inter-frame consistency constraint D1 and the spatial continuity constraint D2:

[0052] y c (i) represents the y-coordinate of the edge point in the current frame. p The y-coordinate of the edge point corresponding to the previous frame. c (i+1) is the ordinate of the edge point in the next frame.

[0053] Set thresholds T1=0.05 pixels and T2=0.2 pixels, retain only edge points that satisfy D1≤T1 and D2≤T2 as true edges, iterate frame by frame to track, remove false edges caused by noise, background structure and sudden changes in lighting, and output a continuous and smooth displacement time history.

[0054] Draw the feature lines of the bar's texture, and extract the RGB color information of the bar's surface. Through color change analysis, identify appearance defects such as rust and wear on the bar's surface.

[0055] The appearance parameters of the rods include RGB chromaticity and texture feature lines, while the structural vibration parameters include the vibration displacement, vibration frequency, frequency-locked range, vibration duration, and reduced wind speed range of the flexible rods. The extraction and processing of these structural vibration parameters integrates sub-pixel edge localization, wavelet denoising, short-time Fourier transform, and time-frequency analysis techniques to improve the accuracy of parameter measurements and environmental adaptability.

[0056] In an optional embodiment, the extraction process of RGB chromaticity and texture feature lines is as follows: The video framing algorithm in OpenCV is called, passing in the video path to be framed and the file path where the images are saved. Vibration information of each frame is obtained from the video file, and keyframes are extracted. Based on histogram differences, the Canny operator for edge detection is used to obtain the shape features of the flexible rod, non-interested regions are cut, background interference is removed, and the RGB chromaticity information of the image is extracted to draw texture feature lines. Finally, a threshold is set according to the 90% chromaticity quantile line among all pixels, and each frame image is binarized, that is, the chromaticity within a certain range is adjusted to 0 or 255 to enhance chromaticity difference recognition and improve the accuracy of rod surface defect recognition.

[0057] The calculation process of the Canny operator is as follows: the image is smoothed using a Gaussian filter, the gradient is calculated based on the first-order partial derivative of the Gaussian function, and the local maximum value of the gradient magnitude is detected; then, a low threshold T1 is used to obtain weak edge E1, and similarly, a high threshold T2 is used to obtain edge E2, and only the connected components that have a connection relationship with E2 are retained as the output edge E in E1 to ensure the accuracy of edge extraction.

[0058] In an optional embodiment, the vibration displacement signal is denoised using wavelet denoising. The signal is transformed using wavelet transform to obtain high-frequency components and approximately low-frequency components. High-frequency noise is thresholded, and the processed components are then reconstructed using wavelet denoising to obtain a smooth and stable denoised displacement time history. Let the noisy signal model be:

[0059] x(t) is the noisy vibration signal, s(t) is the real vibration signal, and n(t) is the noise signal; The coefficients after wavelet transform are as follows:

[0060] W x W represents the wavelet transform coefficients of the noisy signal. sW represents the wavelet transform coefficients of the real signal. n These are the wavelet transform coefficients of the noise signal; Denoising is achieved using a soft thresholding function, expressed as follows:

[0061] x denoised wavelet coefficients, λ is the denoising threshold, and sgn(·) is the sign function.

[0062] Step 3: Based on the aforementioned appearance parameters and structural vibration parameters, scale calibration is performed using the time baseline parallax method and multi-source data fusion method to establish the conversion relationship between pixel displacement and actual displacement. Vibration intensity is calculated based on Hilbert envelope analysis and moving root mean square (RMS) analysis. An adaptive threshold discrimination mechanism is constructed using the vibration-free section as a baseline to automatically identify the initiation and continuation sections of vortex-induced vibration.

[0063] First, establish the mapping relationship between image coordinates and actual physical coordinates. Use the time baseline parallax method to calibrate pixels and actual displacement scales. Measure the physical length L of the rod and read the image pixel length p. L Calculate the scaling factor λ = L / p L To correct the tilt error between the physical plane and the imaging plane, a multi-source data fusion method is employed to fuse the displacement from the accelerometer sensor. The conversion coefficients are then optimized using the least squares method to construct the objective function.

[0064] The optimized conversion coefficients obtained are shown below, which lay the foundation for subsequent displacement correction and vibration parameter calculation.

[0065]

[0066] In the formula, J(α) is the objective function, α is the transformation coefficient, N is the number of data samples, and u i For the accelerometer to measure displacement values, x i Extract displacement values ​​from the image.

[0067] Furthermore, the vortex-induced vibration initiation identification method involves performing a Fast Fourier Transform (FFT) on the calibrated displacement-time history to extract the dominant frequency f0 of the vortex-induced vibration and plot the power spectrum.

[0068] Narrowband bandpass filtering with a value of 0.7f0 to 1.3f0 is used to filter out low-frequency drift and high-frequency noise; Performing a Hilbert transform on the filtered signal yields the analytic signal and the instantaneous envelope, as shown in the formula:

[0069] The sliding RMS is calculated using a window of 5 to 10 times the vibration period; the sliding root mean square RMS is calculated as follows:

[0070] N is the number of data points within the sliding window, Δt is the sampling interval, and x(t-kΔt) is the vibration displacement value at the corresponding time.

[0071] Using data from vibration-free sections as the baseline, the baseline mean μ and standard deviation α are calculated to construct an adaptive vibration initiation threshold:

[0072] In the formula, k is taken as 4; when the sliding RMS continuously exceeds the above threshold and remains for a preset time (e.g., more than 10 cycles), it is determined that vortex-induced vibration has entered.

[0073] Step 4: Use Short Time Fourier Transform (STFT) to perform time-frequency analysis on the displacement signal, and determine the frequency locking characteristics of vortex-induced vibration based on vibration energy and dominant frequency.

[0074] Furthermore, a short-time Fourier transform (STFT) is performed on the vibration displacement signal for time-frequency analysis and frequency locking determination.

[0075] x(τ) is the vibration displacement signal, w(τ-t) is the window function, f is the frequency, and j is the imaginary unit. If the vibration energy is concentrated in a single frequency band and the main frequency fluctuation is <0.3Hz, it is determined to be a frequency-locked state.

[0076] Step 5: Construct a two-dimensional CFD single-degree-of-freedom fluid-structure interaction numerical model based on the structural vibration parameters of the members. Use the SST k-ω turbulence model, dynamic mesh technology and fourth-order Runge-Kutta method to solve the structural motion equations. Simulate the amplitude, frequency and wake vortex-induced evolution under different reduced wind speeds to verify the identification results and vortex-induced vibration mechanism. Assess the risk of vortex-induced vibration and output operation and maintenance recommendations based on the assessment level.

[0077] Based on the extracted vibration parameters (vibration displacement, vibration frequency) of the flexible rod structure, and combined with fluid-structure interaction theory and structural dynamics principles, a two-dimensional single-degree-of-freedom fluid-structure interaction numerical model of the transmission tower's flexible rod is established. The three-dimensional spatial structure of the transmission cylindrical flexible steel pipe rod is equivalent to a two-dimensional crosswind single-degree-of-freedom vibration system, ignoring along-wind vibration and higher-order bending modes, retaining only the first-order dominant mode shape. Based on Euler-Bernoulli beam theory and the modal superposition principle, an amplitude amplification factor is introduced to correct the modal effect, and the structural dynamics equations are established.

[0078] m is the equivalent mass per unit length of the rod, c is the equivalent damping coefficient, and k is the equivalent stiffness coefficient. These represent the crosswind acceleration, velocity, and displacement of the member, respectively, F. L (t) represents the fluid force; The amplitude amplification factor for the first mode shape is taken as [value missing], and for single-plate members [value missing]. .

[0079] The fluid domain uses SST The turbulence model uses a computational domain with a transverse width of 25D and a wake length of 32D, where D is the outer diameter of the rod. A structured mesh is used throughout the domain, with localized mesh refinement in the near-wall region of the cylinder. The first mesh layer near the wall satisfies the dimensionless wall distance. It accurately captures the characteristics of boundary layer separation and Karman vortex-induced shedding. Among them This is a general dimensionless wall distance in CFD, representing the normalized normal distance from the first-layer mesh node near the wall to the bar wall.

[0080] The structural motion equations are solved using a dynamic mesh, a user-defined function (UDF), and the fourth-order Runge-Kutta method. In the formula: This refers to the time step for numerical computation. This is the acceleration slope term; This is the velocity slope term; The velocity and displacement at the current moment; Update velocity and displacement for the next moment; Calculate the reduced wind speed and dimensionless amplitude:

[0081] U represents the actual wind speed, f n Let A be the natural frequency of the rod, D be the diameter of the rod, and A be the diameter of the rod. max For the maximum vibration amplitude, U r To reduce wind speed, A The amplitude is dimensionless.

[0082] The simulation results are compared with the actual measurements to complete the mechanism verification and solve the vortex-induced vibration dynamics information of the rod (vortex shedding frequency, amplitude, cable force, etc.). If the mechanical information exceeds the safety limit of the rod, it is determined that the rod has vibration damage, and the preliminary judgment basis for the damage level is clarified.

[0083] The appearance parameters, structural vibration parameters, and mechanical information of the members are input into a pre-constructed vortex-induced vibration risk assessment model. Weights W1 to W5 are assigned to the structural indicators, corresponding to amplitude, frequency, frequency lock, duration, and wind speed range, respectively. A weighted scoring method is used, with the following scoring formula:

[0084] Gt For structural scoring, n is the number of indicators, W i Assuming the weights of each indicator, B i The judgment value is 0 / 1 (0 for exceeding the limit, 1 for normal), A i A represents the measured value of the indicator. lim For the limit value of the index, A max This represents the maximum value of the indicator.

[0085] The formula is used to calculate the real-time safety rating of the output members, thereby quantifying the risk status of vortex-induced vibration of the flexible members of the transmission tower.

[0086] Using a seven-day monitoring cycle, the monitoring data within the cycle undergoes systematic measurement, wavelet denoising, and time-frequency analysis. Predictions are performed three times daily, and a structure score is calculated. A daily representative value G is taken according to a 3% deviation rule. ti Calculate the deviation δ j =|G tij -G ti2 |, if δ j / G ti2 If the value is less than 3%, the average of the three measurements is taken; otherwise, the median value is taken or the measurement is repeated to determine the representative value of the daily structural score, forming continuous time series data to provide data support for subsequent vibration trend prediction.

[0087] Perform a Spearman correlation test on the time series within the period to determine the stationarity of the data. The correlation coefficient is calculated using the following formula:

[0088] The statistics are constructed as shown below, and the stationarity of the sequence is determined by the statistics.

[0089]

[0090] q s Here, is the Spearman correlation coefficient, n is the data sample size, t is the time series index, and R0 is the time series index. t For G ti The rank of is T, and T is a statistic.

[0091] Perform first-order difference b on the non-stationary sequence t =a （t+1） -a t The ARIMA model was used to predict the structural state score G on day 8. t8 This enables early prediction of the risk of vortex-induced vibration in flexible rods.

[0092] According to the representative value of the score and the predicted score within the monitoring period, calculate the monitoring indicators A (the proportion of scores > 90) and B (the proportion of scores < 80), determine the final risk level by combining the two indicators, divide it into levels I to V, and output the start and end times of vortex-induced vibration, vibration amplitude, main frequency, lock-in frequency range, and risk level. Give maintenance and operation suggestions based on fatigue characteristics: Level I: Normal operation; Level II: Closely monitor; Level III: Restrict the passage of large vehicles; Level IV: Install flow disturbance devices; Level V: Conduct local inspection and reinforcement or suspend the line operation. Combine the risk level and the specific vibration damage situation to formulate targeted maintenance and operation disposal plans and traffic control suggestions to ensure the scientificity and practicality of the control measures.

[0093] After implementing the maintenance and operation disposal plan, monitor the changes in the vibration indicators of the flexible rod in real time, dynamically adjust the parameter weights, thresholds, and edge constraint coefficients in the risk scoring model through a closed-loop feedback mechanism, continuously optimize the recognition and evaluation accuracy, and form a full-process maintenance and operation system for monitoring, recognition, evaluation, disposal, feedback, and optimization. Upload all monitoring data, vibration recognition results, time-frequency spectra, risk assessment reports, and maintenance records to the intelligent maintenance cloud platform to achieve data sharing, real-time early warning of abnormalities, and efficient management of maintenance and operation. At the same time, modify the algorithm parameters according to the feedback of the on-site verification results, update the thresholds, weights, and edge constraint coefficients, and improve the long-term monitoring accuracy and robustness.

[0094] At the same time, encapsulate the entire process into a visualization software, which supports video import, automatic frame division, edge extraction, displacement calculation, vortex-induced vibration recognition, and report export in one click; connect the monitoring data to the intelligent maintenance system of the transmission line to achieve multi-tower linkage and hierarchical early warning. The system provides functions such as user login, data management, historical query, abnormal alarm, and maintenance work order, providing full-process support for the safe operation of transmission towers. Connect to the intelligent maintenance system of the transmission line to achieve multi-tower linkage, big data analysis, trend prediction, and hierarchical early warning, providing comprehensive and accurate decision-making support for the management and maintenance departments of transmission lines.

[0095] Compared with the existing technology, the present invention provides a method and system for identifying the vortex-induced vibration response of flexible rods of transmission towers based on machine vision, which has the following beneficial effects: 1. Compared with the traditional sensor monitoring method, there is no need to deploy a large number of sensors and lines, and the installation and maintenance costs are significantly reduced; compared with the manual inspection method, it can achieve automated and continuous monitoring, is convenient and efficient to operate, and improves the accuracy of vibration recognition and displacement measurement through sub-pixel positioning technology, effectively avoiding the subjective errors of manual observation.

[0096] 2. Through image gray feature extraction and chromaticity analysis, not only can the structural parameters such as the displacement and frequency of the vortex-induced vibration of the rod be identified, but also the chromaticity and texture changes caused by corrosion, wear, and fatigue damage on the surface of the rod can be sensed, realizing the synchronous monitoring of appearance diseases and structural vibration damage, and improving the comprehensiveness of the maintenance and operation of transmission towers.

[0097] 3. No modification to the transmission tower structure is required. Vibration data of flexible members at different locations and of different types can be obtained simply through industrial image acquisition equipment. There is no need to deploy complex measurement systems on site. The measurement efficiency is high, and it can adapt to the needs of long-distance monitoring in complex field environments, reducing the difficulty of on-site operations.

[0098] 4. By adopting the adaptive thresholding method and wavelet denoising technology, the image grayscale threshold and signal filtering parameters can be automatically adjusted to reduce the measurement error caused by changes in outdoor light intensity, uneven lighting and environmental noise, solve the problem of low vibration recognition accuracy under complex working conditions, and improve the environmental adaptability and robustness of the method.

[0099] 5. A multi-index-based vortex-induced vibration risk assessment system was constructed. The system combines core indicators such as vibration amplitude, dominant frequency, and frequency-locking range with weights to reduce the one-sided influence of a single indicator and emphasize the comprehensive judgment of the overall vibration state of the tower members. This system is more operational in engineering practice. At the same time, the system enables hierarchical operation and maintenance and traffic control based on the evaluation results, which can keep the flexible tower members of the transmission tower in a safe operating state.

[0100] 6. Based on the combination of time-frequency analysis and numerical simulation, the vibration data of flexible rods are combined with the verification of fluid-structure interaction mechanism. This not only provides a comprehensive understanding of the vortex-induced vibration characteristics of the rods, but also provides data support for the operation and maintenance decision of transmission towers. This enables vibration monitoring to serve the optimization of power grid safety and helps to promote the construction of intelligent operation and maintenance of transmission lines.

[0101] 7. It has a wide range of applications, not only applicable to vortex-induced vibration monitoring of flexible members of transmission towers, but also applicable to vibration identification of flexible components of tall structures such as wind power towers and communication towers, as well as synchronous monitoring of surface defects and vibration damage of various flexible structures, providing useful support and assistance for promoting intelligent construction and safe operation and maintenance of power infrastructure.

[0102] This invention can be a system, method, and / or computer program product. This invention also discloses a machine vision-based system for identifying the vortex-induced vibration response of flexible transmission tower members, based on the aforementioned machine vision-based method for identifying the vortex-induced vibration response of flexible transmission tower members, comprising: The image preprocessing module is used to acquire the vibration video stream of the flexible pole of the transmission tower using industrial image acquisition equipment, decompose the video stream into an ordered frame sequence, extract the region of interest at the edge of the pole from the frame sequence, perform frame-by-frame magnification using bicubic interpolation, and obtain the subpixel level coordinates of the pole edge based on the subpixel edge localization method. The edge parameter extraction module is used to combine the double threshold constraint edge tracking algorithm to retain the edges that simultaneously satisfy the inter-frame consistency constraint and the spatial continuity constraint as real edges and remove false edges to obtain continuous and stable vibration displacement time history data. Based on the sub-pixel level coordinates, the appearance parameters and structural vibration parameters of the rod are extracted. The vortex-induced vibration identification module is used to perform scale calibration based on the appearance parameters and structural vibration parameters by using the time baseline parallax method and multi-source data fusion method, establish the conversion relationship between pixel displacement and actual displacement, calculate vibration intensity based on Hilbert envelope analysis and sliding root mean square (RMS), and construct an adaptive threshold discrimination mechanism with the vibration-free section as the baseline to automatically identify the vortex-induced vibration initiation and continuous sections. The frequency locking characteristic determination module is used to perform time-frequency analysis on the displacement signal using short-time Fourier transform (STFT) and determine the frequency locking characteristic of vortex-induced vibration based on vibration energy and dominant frequency. The identification result verification module is used to construct a two-dimensional CFD single-degree-of-freedom fluid-structure interaction numerical model based on the structural vibration parameters of the members. It uses the SST k-ω turbulence model, dynamic mesh technology and the fourth-order Runge-Kutta method to solve the structural motion equations, simulate the amplitude, frequency and wake vortex-induced evolution under different reduced wind speeds, verify the identification results and vortex-induced vibration mechanism, assess the risk of vortex-induced vibration, and output operation and maintenance suggestions based on the assessment level.

[0103] Based on the spirit of this invention, those skilled in the art will readily conceive of a computer program product derived from the aforementioned machine vision-based method for identifying the vortex-induced vibration response of flexible transmission tower members. The computer program product may include a computer-readable storage medium on which computer-readable program instructions are loaded to enable a processor to implement various aspects of this disclosure. That is, this application also includes a terminal comprising a processor and a storage medium; the storage medium is used to store instructions; the processor is used to operate according to the instructions to execute the steps of the aforementioned machine vision-based method for identifying the vortex-induced vibration response of flexible transmission tower members.

[0104] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example, but not limited to, electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0105] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0106] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0107] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A machine vision-based method for identifying the vortex-induced vibration response of flexible members in transmission towers, characterized in that, Includes the following steps: Step 1: Use industrial image acquisition equipment to acquire the vibration video stream of the flexible pole of the transmission tower, decompose the video stream into an ordered frame sequence, extract the region of interest at the edge of the pole from the frame sequence, use bicubic interpolation to magnify frame by frame, and obtain the subpixel level coordinates of the pole edge based on the subpixel edge localization method; Step 2: Combining the double threshold constraint edge tracking algorithm, retain the edges that simultaneously satisfy the inter-frame consistency constraint and the spatial continuity constraint as real edges and remove false edges to obtain continuous and stable vibration displacement time history data. Based on the sub-pixel level coordinates, extract the appearance parameters and structural vibration parameters of the rod. Step 3: Based on the appearance parameters and structural vibration parameters, scale calibration is performed using the time baseline parallax method and multi-source data fusion method to establish the conversion relationship between pixel displacement and actual displacement. Vibration intensity is calculated based on Hilbert envelope analysis and sliding root mean square (RMS) analysis. An adaptive threshold discrimination mechanism is constructed with the vibration-free section as the baseline to automatically identify the vortex-induced vibration initiation and continuous sections. Step 4: Use Short Time Fourier Transform (STFT) to perform time-frequency analysis on the displacement signal, and determine the frequency locking characteristics of vortex-induced vibration based on vibration energy and dominant frequency; Step 5: Construct a two-dimensional CFD single-degree-of-freedom fluid-structure interaction numerical model based on the structural vibration parameters of the members. Use the SSTk-ω turbulence model, dynamic mesh technology and fourth-order Runge-Kutta method to solve the structural motion equations. Simulate the amplitude, frequency and wake vortex-induced evolution under different reduced wind speeds to verify the identification results and vortex-induced vibration mechanism. Assess the risk of vortex-induced vibration and output operation and maintenance recommendations based on the assessment level.

2. The method for identifying the vortex-induced vibration response of flexible transmission tower members based on machine vision according to claim 1, characterized in that: In step 1, the step of extracting the region of interest (ROI) of the rod edge from the frame sequence and performing frame-by-frame magnification using bicubic interpolation further includes: Select a rectangular region of interest (ROI) that includes the boundaries of the poles, avoiding interference areas caused by conductors, insulators, and tower intersecting components; apply bicubic interpolation to the ROI region to suppress jagged edges and blurring, while maintaining the continuity of edge grayscale gradients. In the formula, w(i,j) is the bicubic interpolation weight function, I(x,y) is the original ROI image, and I'(x',y') is the magnified image.

3. The method for identifying the vortex-induced vibration response of flexible transmission tower members based on machine vision according to claim 2, characterized in that: In step 1, the method for obtaining sub-pixel level coordinates of the rod edge based on sub-pixel edge localization further includes: Construct a 5×3 local window centered on the target pixel, and calculate the sum of gray levels S in the left, middle, and right columns respectively. L S M S R Sub-pixel localization is achieved by inferring edge line parameters based on the pixel area segmentation model. Where A and B are the grayscale values ​​of the rod and the background, h is the pixel side length, and E is the grayscale value of the background. L E M E R Let a and b be the area enclosed by the edge line within the corresponding column, where a and b are the parameters of the edge line.

4. The method for identifying the vortex-induced vibration response of flexible transmission tower members based on machine vision according to claim 3, characterized in that: Step 2 further includes: Combining a dual-threshold edge tracking algorithm to eliminate false edge interference, calculate the inter-frame consistency constraint D1 and the spatial continuity constraint D2: y c (i) represents the y-coordinate of the edge point in the current frame. p The y-coordinate of the edge point corresponding to the previous frame. c (i+1) is the ordinate of the corresponding edge point in the next frame; Set thresholds T1=0.05 pixels and T2=0.2 pixels, retain only edge points that satisfy D1≤T1 and D2≤T2 as true edges, iterate frame by frame to track, remove false edges caused by noise, background structure and sudden changes in lighting, and output a continuous and smooth displacement time history.

5. The method for identifying the vortex-induced vibration response of flexible transmission tower members based on machine vision according to claim 4, characterized in that: The step of scaling by using the time baseline parallax method and multi-source data fusion method to establish the conversion relationship between pixel displacement and actual displacement further includes: A mapping relationship between image coordinates and actual physical coordinates is established. The time baseline parallax method is used for pixel and actual displacement scale calibration. The physical length L of the rod is measured, and the image pixel length p is read. L Calculate the scaling factor λ = L / p L To correct the tilt error between the physical plane and the imaging plane, a multi-source data fusion method is employed to fuse the displacement from the accelerometer sensor. The conversion coefficients are then optimized using the least squares method to construct the objective function. The optimized conversion coefficients are obtained by solving: In the formula, J(α) is the objective function, α is the transformation coefficient, N is the number of data samples, and u i For the accelerometer to measure displacement values, x i Extract displacement values ​​from the image.

6. The machine vision-based method for identifying the vortex-induced vibration response of flexible transmission tower members as described in claim 5, characterized in that, The vibration intensity calculation based on Hilbert envelope analysis and sliding root mean square (RMS) further includes an adaptive threshold discrimination mechanism that uses the vibration-free section as a baseline to automatically identify the initiation and continuation sections of vortex-induced vibration. Perform a Fast Fourier Transform (FFT) on the calibrated displacement-time history to extract the dominant frequency f0 of the vortex-induced vibration and plot the power spectrum. Narrowband bandpass filtering with a value of 0.7f0 to 1.3f0 is used to filter out low-frequency drift and high-frequency noise; Performing a Hilbert transform on the filtered signal yields the analytic signal and the instantaneous envelope: The sliding RMS is calculated using a window of 5 to 10 times the vibration period; the sliding root mean square RMS is calculated as follows: N is the number of data points within the sliding window, Δt is the sampling interval, and x(t-kΔt) is the vibration displacement value at the corresponding time. Using data from vibration-free sections as the baseline, the baseline mean μ and standard deviation α are calculated to construct an adaptive vibration initiation threshold: In the formula, k is taken as 4; when the sliding RMS continuously exceeds the above threshold and remains for a preset time, it is determined that vortex-induced vibration has entered.

7. The method for identifying the vortex-induced vibration response of flexible transmission tower members based on machine vision according to claim 6, characterized in that: Step 4 further includes: The vibration displacement signal is subjected to a short-time Fourier transform (STFT) for time-frequency analysis and frequency locking determination. x(τ) is the vibration displacement signal, w(τ-t) is the window function, f is the frequency, and j is the imaginary unit. If the vibration energy is concentrated in a single frequency band and the main frequency fluctuation is <0.3Hz, it is determined to be a frequency-locked state.

8. The method for identifying the vortex-induced vibration response of flexible transmission tower members based on machine vision according to claim 7, characterized in that: Step 5 further includes: Based on the extracted vibration parameters of the flexible rod structure, and combining fluid-structure interaction theory and structural dynamics principles, a two-dimensional single-degree-of-freedom fluid-structure interaction numerical model of the transmission tower's flexible rod is established. The three-dimensional spatial structure of the transmission cylindrical flexible steel pipe rod is equivalent to a two-dimensional crosswind single-degree-of-freedom vibration system, ignoring along-wind vibration and higher-order bending modes, retaining only the first-order dominant mode shape. Based on Euler-Bernoulli beam theory and the modal superposition principle, an amplitude amplification factor is introduced to correct the modal effect, and the structural dynamics equations are established. m is the equivalent mass per unit length of the rod, c is the equivalent damping coefficient, and k is the equivalent stiffness coefficient. These represent the crosswind acceleration, velocity, and displacement of the member, respectively, F. L (t) represents the fluid force; This is the amplitude amplification factor for the first mode shape; The fluid domain uses SST The turbulence model uses a computational domain with a transverse width of 25D and a wake length of 32D, where D is the outer diameter of the rod. A structured mesh is used throughout the domain, with localized mesh refinement in the near-wall region of the cylinder. The first mesh layer near the wall satisfies the dimensionless wall distance. To capture the characteristics of boundary layer separation and Karman vortex-induced shedding; among which This is a general dimensionless wall distance in CFD, representing the normalized normal distance from the first-layer mesh node near the wall to the bar wall. The structural motion equations are solved using a dynamic mesh, a user-defined function (UDF), and the fourth-order Runge-Kutta method. In the formula: This refers to the time step for numerical computation. This is the acceleration slope term; This is the velocity slope term; The velocity and displacement at the current moment; Update velocity and displacement for the next moment; Calculate the reduced wind speed and dimensionless amplitude: U represents the actual wind speed, f n Let A be the natural frequency of the rod, D be the diameter of the rod, and A be the diameter of the rod. max For the maximum vibration amplitude, U r To reduce wind speed, A The amplitude is dimensionless.

9. The method for identifying the vortex-induced vibration response of flexible transmission tower members based on machine vision according to claim 8, characterized in that: Step 5 further includes: Output the start and end times of eddy-induced vibration, vibration amplitude, dominant frequency, frequency locking range, and hazard level; and provide graded operation and maintenance suggestions based on fatigue characteristics, including close monitoring, restricting the passage of large vehicles, installing turbulence devices, local maintenance and reinforcement, or line shutdown.

10. A machine vision-based system for identifying the vortex-induced vibration response of flexible members in transmission towers, characterized in that, include: The image preprocessing module is used to acquire the vibration video stream of the flexible pole of the transmission tower using industrial image acquisition equipment, decompose the video stream into an ordered frame sequence, extract the region of interest at the edge of the pole from the frame sequence, perform frame-by-frame magnification using bicubic interpolation, and obtain the subpixel level coordinates of the pole edge based on the subpixel edge localization method. The edge parameter extraction module is used to combine the double threshold constraint edge tracking algorithm to retain the edges that simultaneously satisfy the inter-frame consistency constraint and the spatial continuity constraint as real edges and remove false edges to obtain continuous and stable vibration displacement time history data. Based on the sub-pixel level coordinates, the appearance parameters and structural vibration parameters of the rod are extracted. The vortex-induced vibration identification module is used to perform scale calibration based on the appearance parameters and structural vibration parameters by using the time baseline parallax method and multi-source data fusion method, establish the conversion relationship between pixel displacement and actual displacement, calculate vibration intensity based on Hilbert envelope analysis and sliding root mean square (RMS), and construct an adaptive threshold discrimination mechanism with the vibration-free section as the baseline to automatically identify the vortex-induced vibration initiation and continuous sections. The frequency locking characteristic determination module is used to perform time-frequency analysis on the displacement signal using short-time Fourier transform (STFT) and determine the frequency locking characteristic of vortex-induced vibration based on vibration energy and dominant frequency. The identification result verification module is used to construct a two-dimensional CFD single-degree-of-freedom fluid-structure interaction numerical model based on the structural vibration parameters of the members. It uses the SST k-ω turbulence model, dynamic mesh technology and the fourth-order Runge-Kutta method to solve the structural motion equations, simulate the amplitude, frequency and wake vortex-induced evolution under different reduced wind speeds, verify the identification results and vortex-induced vibration mechanism, assess the risk of vortex-induced vibration, and output operation and maintenance suggestions based on the assessment level.

11. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the machine vision-based method for identifying the vortex-induced vibration response of flexible transmission tower members according to any one of claims 1-9.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the machine vision-based method for identifying the vortex-induced vibration response of flexible members of transmission towers as described in any one of claims 1-9.

Images