A method for monitoring bridge surface defects and structural damage based on mobile app and machine vision

Abstract

This invention discloses a method for monitoring bridge surface defects and structural damage based on a mobile app and machine vision. It utilizes mobile devices to acquire bridge appearance and structural data; processes videos and raw images on an Android platform; establishes a data model based on the bridge images; establishes a mechanical model based on the images and structural data to obtain bridge mechanical information; compares structural and color data with an initial database; identifies defect locations and bridge damage conditions; and shares information and provides feedback on maintenance plans. This invention, through big data analysis of bridge monitoring, extends traditional sensor methods to mobile devices, significantly reducing the installation and maintenance costs of monitoring equipment. Simultaneously, it establishes a mechanical model, compares existing bridge structural data with past data, constructs an evaluation system for different structural indicators, and integrates traffic data with bridge structural data. A comprehensive evaluation is then used to establish an effective computational model, generating bridge maintenance and management plans.

Description

Technical Field

[0001] This invention relates to a method for monitoring bridge surface defects and structural damage based on mobile app and machine vision, belonging to the fields of intelligent construction and traffic management. Background Technology

[0002] With the continuous advancement of China's modernization and the increasing level of intelligent construction, cities are expanding exponentially in terms of area, population, and vehicles, leading to increasingly busy urban traffic. The passage of large vehicles and traffic congestion during rush hours pose significant safety hazards to urban transportation. Bridges, as a crucial component of modern urban transportation, frequently connect residents on both sides of rivers and bear heavy traffic loads. Their structural safety urgently requires the intervention of intelligent monitoring technology for targeted solutions. Once a bridge experiences an accident or faces a serious threat, it can range from affecting the smooth flow of urban traffic to causing localized traffic paralysis, resulting in significant loss of life and property, and impacting the city's image and development. On May 5, 2020, the Humen Bridge in Guangdong experienced sudden vibrations, causing traffic disruptions and resulting in direct and indirect losses, including daily toll fees, amounting to tens of millions of yuan. Furthermore, in cities with well-developed water systems (such as Suzhou and Wuxi), bridges play a pivotal role in urban transportation. The impact of bridge damage and surface defects is incalculable. Therefore, research on the health monitoring of bridge structural damage and surface defects in urban transportation is urgently needed.

[0003] Actual research reveals that bridge health monitoring systems, due to their high cost, are mostly concentrated on extra-large bridges (such as the Nanjing Yangtze River Bridge and the Sutong Bridge). Statistics show that approximately 67% of large bridges over 300 meters (including cable-stayed bridges and suspension bridges) have installed bridge health monitoring systems. However, for most small and medium-sized bridges, due to design costs and the difficulty of sensor installation and maintenance, no safety monitoring measures have been implemented, posing significant safety hazards. Data shows that out of more than 260 small and medium-sized bridges, only 0.04‰ have installed bridge health monitoring systems. Urban bridges are numerous and diverse, and many bridges cannot afford the high cost of health monitoring systems, relying instead on periodic inspections and manual patrols, or repairs only when major problems occur. As new bridges are continuously built, the problems of aging bridges become increasingly apparent, and these outdated maintenance methods place enormous workload on municipal bridge management departments.

[0004] Because China's bridge modernization started relatively late, the structural strength of most bridges remains within a reasonable range, and structural fatigue has not yet occurred. However, with the passage of time, health problems in bridge structures will become more prominent. Currently, approximately 40% of bridges have been in service for over 20 years, and more than 100,000 bridges are considered dangerous, posing significant safety hazards. Suitable technical solutions for addressing these issues have yet to be found. However, with the development of the internet age, the rapid advancement of artificial intelligence (represented by ChatGPT), and mobile phone camera resolutions reaching 50 megapixels and computing power reaching 7 TOPS (7 trillion operations per second), mobile devices offer advantages such as high efficiency and low cost for safety monitoring. With the increasing prevalence of mobile devices, real-time data collection and timely solutions can be achieved. This invention, based on the widespread use of mobile devices in the internet age, combines artificial intelligence machine vision technologies used in bridge monitoring systems with mobile devices, proposing an effective and feasible solution for monitoring small and medium-sized bridges. Based on this invention, combining digital image processing with mobile devices will significantly reduce manpower and material costs. Summary of the Invention

[0005] This invention provides a method for monitoring bridge surface defects and structural damage based on a mobile app and machine vision, which solves the problems disclosed in the background art.

[0006] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows:

[0007] A method for monitoring bridge surface defects and structural damage based on mobile app and machine vision:

[0008] Using mobile devices to capture video images of bridges, the appearance and structural parameters of the bridges are extracted from the video images;

[0009] Establish a three-dimensional spatial coordinate system on the bridge deck, obtain the coordinates of each position of the bridge, and merge the bridge position coordinates with the bridge appearance parameters to obtain the appearance parameters corresponding to each position.

[0010] The appearance parameters corresponding to each location are compared with the initial appearance parameters of the bridge. If the difference between the comparison at that location exceeds the threshold, it indicates that there is a defect at that location, and the location of the defect is displayed on the mobile app.

[0011] A bridge mechanical model is constructed based on the bridge's structural parameters. The mechanical information of the bridge is obtained through the mechanical model. If the mechanical information of the bridge exceeds the safety limit of the bridge itself, it indicates that there is damage to the bridge structure.

[0012] The bridge's appearance parameters, structural parameters, and mechanical information are input into a pre-built structural scoring model, which outputs the bridge's safety rating.

[0013] Furthermore, the appearance parameters of the bridge include RGB chromaticity and texture feature lines; the structural parameters of the bridge include the displacement and vibration frequency of the bridge deck, and the displacement and vibration frequency of the cables.

[0014] Furthermore, the extraction process of RGB chromaticity and texture feature lines is as follows:

[0015] The OpenCV video framing algorithm is called, taking the video path and the file path where the images are saved as parameters. 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 bridge deck and cables, non-interest areas are clipped, and the RGB chromaticity information of the image is extracted and feature lines are drawn. Finally, a threshold is set according to the 90% chromaticity quantile line of all pixels, and each frame is binarized, that is, the chromaticity within a certain range is adjusted to 0 or 255.

[0016] Furthermore, the computation process of the Canny operator is as follows:

[0017] The image is smoothed using a Gaussian filter. The gradient is calculated based on the first-order partial derivative of the Gaussian function, and local maxima of the gradient magnitude are detected. Then, a low threshold T1 is used to obtain weak edges E1, and similarly, a high threshold T2 is used to obtain edges E2. Only connected components that are connected to E2 are retained as output edges E in E1.

[0018] Furthermore, the displacement extraction process for the bridge deck and cables is as follows: During video framing, an image stabilization algorithm based on mobile phone accelerometer data extraction is used. By capturing and identifying fixed points in the image, the relative displacements of the bridge deck, cables, and fixed points are obtained, and relative vibration data is acquired. This data is then used to correct the image displacement and reduce displacement errors caused by mobile device shaking. The displacement data of the bridge and cables is subtracted from the data of the fixed points to obtain the actual vibrations of the bridge and cables relative to the ground. Finally, the acceleration data from the Orion-CC application is extracted, based on T=4ml. 2 (Δf) 2 , and obtain the corrected relative displacement data.

[0019] Furthermore, noise reduction processing is performed on the displacement and vibration frequencies of the bridge deck and the cables. The process is as follows:

[0020] Wavelet denoising is used to perform wavelet transform on the noise points in the data to obtain the high-frequency values ​​and approximate low-frequency values ​​of each noise point. Thresholding is performed on the high-frequency values, and wavelet reconstruction is performed on each frequency value component after processing to obtain the denoised image.

[0021] Let the noisy image model a(i,j) be:

[0022] a(i,j)=b(i,j)+c(i,j)

[0023] In the formula, a(i,j) is the noisy image after it has been corrupted by noise, b(i,j) is the initial noise-free image, and c(i,j) is Gaussian noise that follows a Gaussian normal distribution.

[0024] After performing wavelet transform on both sides of the equation, we obtain the coefficients after wavelet transform:

[0025] W a =W b +W c

[0026] W a W represents the coefficients of the original image after wavelet transform. b W represents the coefficients of the original noise-free image after wavelet transform. c The coefficients are the Gaussian white noise after forearm transformation;

[0027] W b =W c -W a

[0028] In wavelet transform-based thresholding denoising methods, the threshold function can be selected from the following two options:

[0029] The hard threshold function is shown in equation (1), W a These are the wavelet coefficients after thresholding, where λ is the threshold value. When the absolute value of a wavelet coefficient is less than the given threshold, the value is set to 0; when the wavelet coefficient is greater than the threshold, the original value is retained, resulting in the denoised wavelet coefficients.

[0030]

[0031] The soft thresholding function, as shown in formula (2), sets the value to 0 when the absolute value of the wavelet coefficient is less than the given threshold, and subtracts the threshold from the value of the wavelet coefficient when the wavelet coefficient is greater than the threshold, thus obtaining the denoised wavelet coefficients.

[0032]

[0033] An adaptive threshold selection method is adopted, which dynamically adjusts the threshold and output value according to different wavelet coefficients, as shown in formula (3):

[0034]

[0035] P is a regulating factor. This represents the wavelet coefficients after adaptive thresholding, where T is the threshold set by the function.

[0036] Furthermore, the bridge mechanics model includes a bridge cable mechanics model and a bridge deck mechanics model.

[0037] Furthermore, the process of constructing the bridge cable mechanical model is as follows:

[0038] When the bridge boundary constraints are simply supported, considering the effects of bending stiffness and the variation in suspension bridge length, and based on actual calculations of the cable forces in a certain tied-arch bridge with under-deck construction, the differential equation for cable vibration is derived using string vibration theory without considering the effect of sag. Where EI is the bending stiffness, T is the cable force, m is the mass per unit length of the cable, and y(x,t) is the displacement function of each point on the cable as a function of time. Solving this equation yields...

[0039] Y(X)=Asin(αx)+Bcos(αx)+Csinh(βx)+Dcosh(βx)

[0040] Where A, B, C, and D are constant terms, ω is the natural frequency, and α = [(p 4 +q 4 ) 0.5 -p 2 ] 0.5 ;β=[(p 4 +q 4 ) 0.5 +p 2 ] 0.5 ;p 2 =T / (2EI); q 4 =ρω 2 / (2EI); The following formula can be obtained: Formula considering the influence of bending stiffness:

[0041] T = 4 mL 2 f 2 (n) / n 2 -EI(nΠ / L) 2

[0042] Formula without considering the effect of bending stiffness:

[0043] T = 4ρmL 2 f 2 (n) / n 2

[0044] In the formula, T is the cable tension; m is the cable density; L is the calculated length of the cable; f(n) is the nth natural frequency of the cable; and n is the order of the natural frequency. Taking into account the influence of bending stiffness and length increase, a correction coefficient θ is introduced to modify the calculation formula, resulting in a cable force calculation formula suitable for short suspenders. The fitting function formula is: T=θ×4mL 2 f2 (n) / n 2 By solving the equation and introducing correction coefficients that consider the changes in bending stiffness and bridge length, the cable force, natural frequency, and calculated length of the completed bridge are substituted into the above formula for regression calculation. The calculated value is θ = 0.962. Therefore, the corrected formula for calculating the short suspension rod is: T = 3.848mL. 2 f 2 (n) / n 2 The formula for calculating cable force, taking into account the effects of bending stiffness and the increase in rod length, is as follows:

[0045] T = 3.848 mL 2 f 2 (n) / n 2 L<15m

[0046]

[0047] In the formula, △L is the increase in the length of the hanger under constant load, L1=L+△L.

[0048] Furthermore, the process of constructing the bridge deck mechanical model is as follows:

[0049] Calculate the relevant mechanical data for the entire bridge. Assume the cable inclination angle is θ, the edge distance is l1, the middle distance is l2, and the spacing between the cable bases is a. All cables are under tension. Recorded as negative, F NK Let be the axial force from the k-th cable to the (k+1)-th cable, then:

[0050]

[0051] Will Let M(x) be positive. k Let x be the bending moment at a distance x from the origin in the k-th segment. The segment from the first cable to the second cable is the first segment, and so on. In particular, the k-th cable... The root line reaches point B as the first Section, from point B to The root is the first The same principle applies to subsequent paragraphs, with the middle paragraphs drawn separately; among them

[0052]

[0053] Similarly, the bending moment of the lower half and the bending moment of the middle half can be obtained as follows:

[0054]

[0055] According to the formula Get |σ max |=σ tmax ≤[σ t ];|σmin |=σ cmax ≤[σ c ].

[0056] Furthermore, the scoring method for the structural scoring model is as follows:

[0057] The indicator i is determined by whether it exceeds the threshold (0 / 1), indicating that the indicator i is within the safety limit. Judgment value B of indicator i i A value of 1 indicates the indicator is in a normal state; indicator i exceeds the 90% safety limit. Judgment value B of indicator i i A value of 0 indicates that the corresponding structural parameter is in an abnormal state; simultaneously, the weighted average score of the bridge structure is calculated based on the parameter weights to determine the damage level of the structure; the structural score calculation is shown in formula (4).

[0058]

[0059] The judgment value matrix after indicator 0 / 1 is represented as follows:

[0060]

[0061] Among them, G t Represents the structural score, A i The actual measured value of the representative indicator. B represents the 90% safety expectation limit of the indicator design. i W represents the judgment value of each indicator. i The values ​​represent the weights of each indicator, q = 1, 2, 3, 4, 5, 6. These represent the cable vibration frequency, cable force, bridge deck displacement, bridge deck vibration frequency, chromaticity change, and cable displacement, respectively. The structural indicator weights are obtained according to bridge inspection and monitoring standards. The structural score is located in the range [0, 100], and the higher the score, the safer the bridge structure is overall.

[0062] Furthermore, this monitoring method uses a seven-day monitoring period, and measures, processes, and statistically analyzes the data within the period. Three predicted values ​​are calculated each day, and structural scores are calculated for each of the three measurements to obtain a vector containing the structural scores from the three measurements for day i. We select the structural score representative value G on day i. ti As the true score value of the structure on day i, the representative score value G of the structure on day i is... ti The calculation process is as follows:

[0063] 1. and respectively with Perform the difference. when When the condition is constant, select the cubic structural score. The arithmetic mean of the values ​​is used as the representative structural score G for day i. ti .

[0064] 2. When there exists a set j such that The other set of j satisfies When, take the median value. G, representing the structural score on day i. ti .

[0065] 3. When If the condition is not met, the vector of the structural score for that group of measurements should not be used as the basis for evaluation, and three more measurements need to be performed. Based on the above method, seven representative values ​​of structural scores [G] can be obtained within a monitoring period. t1 G t2 G t3 G t4 G t5 G t6 G t7 Since indicators such as cable vibration frequency, cable sag, and bridge deck displacement are related to time variations, a time series method can be used to establish a predictive model based on data from the eighth day outside the monitoring period. Let the original time series data be denoted as G. ti (i = 1, 2, ..., 7) To prevent spurious regression problems caused by non-stationary data, we first test the stationarity of the data.

[0066] For a given time series sample G ti (i=1,2,…,7), denoted by G ti The rank of R is t =R(G ti Consider the variable pair (t, R). t The Spearman correlation coefficient q for t = 1, 2, ..., n s ,have:

[0067]

[0068] Constructing statistics:

[0069]

[0070] If the time series is non-stationary, in order to construct a stationary series, we can adjust the time series G. ti Perform a first-order difference operation b on (i = 1, 2, ..., 7) t =a t+1 -a t A stable sequence b is obtained. tFinally, the stable sequence is decomposed into long-term trend variation, cyclical variation, and irregular variation. A product algorithm is used, setting the final index variation as: Long-term trend variation × Cyclical variation × Irregular variation. A regression-moving average algorithm is then employed, combining the autoregressive and moving average processes to generate an iterative equation, yielding the representative structural score G for the eighth day. t8 .

[0071] Furthermore, using the above method, a set of predicted data including the second day of the monitoring period and the period outside the monitoring cycle was obtained, resulting in eight sets of representative structural score values. Monitoring indicators A and B can be obtained from these eight sets of representative structural score values. The calculation methods for monitoring indicators A and B are as follows: The eight sets of data are filtered, and the representative structural score value G is calculated. ti The percentage of scores greater than 90 is denoted as monitoring indicator A; the representative value G of the structured score is calculated. ti The percentage of scores below 80 is designated as monitoring indicator B. The higher the hazard level, the more dangerous the structure. Different hazard levels are then assigned, and maintenance and traffic control plans are determined based on these hazard levels and structural parameter scores.

[0072] The beneficial effects achieved by this invention are as follows:

[0073] 1. This invention uses a machine vision method based on mobile devices to acquire internal deformation data of bridge structures. Compared with traditional sensor methods, this method significantly reduces installation and maintenance costs; compared with manual methods, this method is convenient, fast, and highly accurate.

[0074] 2. This invention acquires surface information of an object through a mobile device, vectorizes the external texture to generate chromaticity data, and can perceive chromaticity changes in the appearance of an object's surface caused by defects and oxidation.

[0075] 3. This invention does not require the laying of a large number of lines. It only needs to acquire deformation data of different structures and positions of the bridge through the camera built into the mobile device, which has high measurement efficiency.

[0076] 4. This invention is based on an adaptive thresholding method, which can automatically select the image grayscale threshold, reduce the color error caused by changes in light intensity, and solve the recognition problem caused by different light intensities and uneven lighting conditions.

[0077] 5. This invention constructs an evaluation system with different structural indicators, weakens the control of a single structural indicator, and emphasizes the determination of the overall structural state. It has better operability in engineering practice. At the same time, by using this evaluation system to implement graded traffic flow control, the bridge structure can be kept in a healthy state.

[0078] 6. This invention combines traffic flow and structural data under various conditions based on deep learning methods, which can not only comprehensively grasp the bridge status, but also enable structural status assessment to serve urban traffic optimization, thus helping to promote the construction of intelligent urban transportation.

[0079] 7. This invention has a wide range of applications and can be applied to situations where submerged tunnels, roads, and other structures experience structural vibrations and surface defects, providing valuable support and assistance for promoting intelligent construction. Attached Figure Description

[0080] Figure 1 This is a schematic diagram of the process of the present invention;

[0081] Figure 2 This is a schematic diagram of bridge data acquisition in this invention;

[0082] Figure 3 This is a schematic diagram of the marker point setting in this invention;

[0083] Figure 4 This is a schematic diagram of the binarized image in this invention;

[0084] Figure 5 This invention relates to an image denoising technique based on wavelet transform.

[0085] Figure 6 This is a schematic diagram of bridge mechanics modeling in this invention;

[0086] Figure 7 This is a flowchart of colorimetric recognition and disease location in this invention;

[0087] Figure 8 This is a schematic diagram of the structural condition assessment and maintenance control steps in this invention. Detailed Implementation

[0088] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0089] like Figure 1 As shown, this invention discloses a method for monitoring bridge defects and structural damage based on a mobile app and machine vision, comprising the following steps:

[0090] Step one involves dividing bridge data collection into bridge appearance data collection and bridge structure data collection, such as... Figure 2 As shown, video images were obtained using the camera function of a mobile device. The texture of the bridge's exterior was vectorized and stored in a video file. Subsequently, the RGB color values ​​and texture feature lines of the bridge surface, as well as the displacement and vibration frequency of the bridge deck and the displacement and vibration frequency of the cables, were extracted from the file.

[0091] Step two: During the recording process, adjust the mobile device's focus and white balance, place the target object within the shooting range on the screen, and select the ROI (Region of Interest). At the end of the recording, turn off the camera and crop the beginning and end of the video.

[0092] Step three involves marking detection points in the area of ​​interest on the bridge to enhance the detection effect. This can be done by spraying reflective material or attaching reflective patches, such as... Figure 3 As shown. Fixed rigid bodies other than the bridge within the shooting range are selected as fixed points, which are then used as the origin of the image stabilization algorithm.

[0093] Step four involves calling the video framing algorithm in OpenCV, passing in the video path and the file path where the images are saved. The algorithm retrieves vibration information from each frame of the video file and extracts the keyframes. Based on histogram differences, the Canny operator for edge detection is used to obtain the shape features of the bridge deck and cables. Regions of interest are cropped, and the RGB chromaticity information of the image is extracted, drawing feature lines. Finally, a threshold is set based on the 90% chromaticity quantile of all pixels, and each frame is binarized, adjusting the chromaticity within a certain range to 0 or 255. The binarized image is shown below. Figure 4 As shown. The calculation process of the Canny operator is as follows:

[0094] First, a Gaussian filter is used to smooth the image. The gradient is calculated based on the first-order partial derivative of the Gaussian function, and local maxima of the gradient magnitude are detected. Then, a low threshold T1 is used to obtain weak edges E1, and similarly, a high threshold T2 is used to obtain edges E2. Only connected components that are connected to E2 are retained as output edges E in E1.

[0095] Step 5: During video frame segmentation, an image stabilization algorithm based on mobile phone accelerometer data extraction is used. By capturing and identifying fixed points in the image, the relative displacements of the bridge deck, cables, and fixed points are obtained, yielding relative vibration data. This data is then used to correct image displacement and reduce displacement errors caused by mobile device jitter. According to Galileo's principle of relativity of motion, the equations of motion in an inertial frame remain invariant under Galilean transformation. The jitter in the original data includes the bridge's jitter relative to the camera device plus the camera device's jitter relative to the ground. Since the fixed points are stationary relative to the ground, only the displacement data of the bridge and cables needs to be subtracted from the fixed point data to obtain the actual vibration of the bridge and cables relative to the ground. Finally, the acceleration data from the application "Orion-CC" is extracted, based on T=4ml... 2 (Δf) 2 , and obtain the corrected relative displacement data.

[0096] Step six involves denoising the vibration displacement data of the bridge deck and cables. Wavelet denoising is used to perform wavelet transform on the noise points in the data, obtaining the high-frequency and approximate low-frequency values ​​of each noise point. Thresholding is applied to the high-frequency values, and wavelet reconstruction is then performed on each frequency component to obtain the denoised image. Wavelet transform-based image denoising techniques are as follows: Figure 5 As shown. The specific model is as follows:

[0097] Let the noisy image model a(i,j) be:

[0098] a(i,j)=b(i,j)+c(i,j)

[0099] In the formula, a(i,j) is the noisy image after it has been corrupted by noise, b(i,j) is the initial noise-free image, and c(i,j) is Gaussian noise that follows a Gaussian normal distribution.

[0100] After performing wavelet transform on both sides of the equation, we obtain the coefficients after wavelet transform:

[0101] W a =W b +W c

[0102] W a W represents the coefficients of the original image after wavelet transform. b W represents the coefficients of the original noise-free image after wavelet transform. c The coefficients are the Gaussian white noise after forearm transformation.

[0103] W b =W c -W a

[0104] In wavelet transform-based thresholding denoising methods, the threshold function is mainly selected in the following two ways:

[0105] (1) Hard threshold function, the function is shown in formula (1). W a These are the wavelet coefficients after thresholding, where λ is the threshold. When the absolute value of the wavelet coefficient is less than the given threshold, the value is set to 0; when the wavelet coefficient is greater than the threshold, the original value is retained. This yields the denoised wavelet coefficients.

[0106]

[0107] (2) Soft threshold function, the function is shown in formula (2). When the absolute value of the wavelet coefficient is less than the given threshold, the value is set to 0. When the wavelet coefficient is greater than the threshold, the value of the wavelet coefficient is subtracted from the threshold to obtain the denoised wavelet coefficient.

[0108]

[0109] The present invention adopts an adaptive threshold selection method, which can dynamically adjust the threshold and output value according to different wavelet coefficients, as shown in formula (3).

[0110]

[0111] P is a regulating factor. This represents the wavelet coefficients after adaptive thresholding, where T is the threshold set by the function. This function can overcome the shortcomings of the two denoising functions mentioned above, effectively mitigating the oscillation effect caused by the direct truncation of the threshold function.

[0112] Step 7: When binarizing the image, an adaptive thresholding method is used to divide the image into multiple blocks. A threshold is calculated for each block individually, and the blocks are further subdivided using the calculated thresholds. This increases the threshold for bright areas and decreases the threshold for dark areas, reducing chromaticity errors caused by changes in light intensity. Based on the brightness distribution of different regions of the image, local thresholds are calculated to maximize the variance between segmented images, minimize the intra-class variance, and maximize the inter-class variance. Total variance = intra-class variance + inter-class variance, as shown in the following formula:

[0113] b = ω0ω1(μ0-μ1) 2

[0114] In the formula, b is the inter-class variance; ω0 is the proportion of foreground pixels to the total number of pixels in the image; ω1 is the proportion of background pixels to the total number of pixels in the image; μ0 is the average proportion of all foreground pixels; and μ1 is the average proportion of all background pixels.

[0115] Step 8: Extract the chromaticity texture distribution of the bridge and cables from the denoised image to obtain texture feature lines. The chromaticity difference of the texture feature lines is no more than 5%. Then, generate a chromaticity sample set, remove duplicates from the chromaticity sample set, remove line segments with similar features, and finally mark the intersection of the feature lines to generate a real-time image and integrate it into a real-time image library.

[0116] Step 9: Establish a three-dimensional spatial coordinate system on the bridge deck. (Based on the spatial location of the bridge structure, determine the specific values ​​of the bridge Path and Row, ensuring that each pair of Path and Row corresponds one-to-one with the specific location of the bridge, obtaining a data set covering the bridge structure. Call the Pandas library to find and clean duplicate spatial data, and use GIS technology to vectorize the data. The vector data forms the spatial coordinates of the bridge structure, thereby strengthening the correspondence between location and coordinates and obtaining a clear and intuitive accurate location data model.)

[0117] Step 10: Use the BufferedImage tool built into OpenCV to obtain the RGB chromaticity information of each pixel in the image. Use the addAll() method to merge the bridge position coordinates (x,y) with the RGB chromaticity information (R,G,B) to obtain the position-chromaticity vector (x,y,R,G,B).

[0118] Step 11: Subtract the bridge deck monitoring point location data frame by frame from the data obtained in Step 1 to obtain the displacement data of a single frame image. Establish a relationship between this displacement data and the single-frame vibration time t to obtain the vibration velocity information of the monitoring point at different periods, and thus obtain the vibration power. Perform a Fourier transform on the bridge deck displacement data. Obtain the bridge deck vibration frequency and plot the power spectrum. Determine the first and second order vibration frequencies of the bridge to obtain the bridge vibration shaping diagram.

[0119] Step 12: Based on the mathematical model of the bridge image, the spatiotemporal coordinates of the detection points in the mathematical model, the position-chromaticity vector of the detection points, and the frequency-power relationship information of the structural vibration obtained by Fourier transforming the displacement data are used to draw the power spectrum.

[0120] Step thirteen: Establish a bridge cable mechanical model based on the obtained and processed data. A schematic diagram of the bridge mechanical model is shown below. Figure 6 As shown. When the bridge boundary constraints are simply supported, considering the influence of bending stiffness and the variation in suspension bridge length, based on actual calculations of the cable forces of a certain tied-arch bridge, and without considering the influence of sag, the differential equation of cable vibration is derived according to the string vibration theory. Where EI is the bending stiffness, T is the cable force, m is the mass per unit length of the cable, and y(x,t) is the displacement function of each point on the cable as a function of time. Solving this equation yields...

[0121] Y(X)=Asin(αx)+Bcos(αx)+Csinh(βx)+Dcosh(βx)

[0122] Where A, B, C, and D are constant terms, and ω is the natural frequency. α=[(p 4 +q 4 ) 0.5 -p 2 ] 0.5 ;β=[(p 4 +q 4 ) 0.5 +p 2 ] 0.5 ;p 2 =T / (2EI); q 4 =ρω 2 / (2EI); the following formula can be obtained: (1) Formula considering the influence of bending stiffness:

[0123] T = 4 mL 2 f 2 (n) / n 2 -EI(nΠ / L) 2

[0124] (2) Formula without considering the influence of bending stiffness:

[0125] T = 4ρmL 2 f 2 (n) / n 2

[0126] In the formula, T is the cable tension; m is the cable density; L is the calculated length of the cable; f(n) is the nth natural frequency of the cable; and n is the order of the natural frequency. Considering the effects of bending stiffness and length increase, a correction factor θ is introduced to modify the calculation formula, resulting in a suitable cable force calculation formula for short suspenders. The fitted function formula is: T = θ × 4mL 2 f 2 (n) / n 2 By solving the equation and introducing correction coefficients that consider the changes in bending stiffness and bridge length, the cable force, natural frequency, and calculated length of the completed bridge are substituted into the above formula for regression calculation. The calculated value is θ = 0.962. Therefore, the corrected formula for calculating the short suspension rod is: T = 3.848mL. 2 f 2 (n) / n 2 The formula for calculating cable force, taking into account both bending stiffness and the effect of the increase in suspender length, is as follows:

[0127] T = 3.848 mL 2 f 2 (n) / n 2 L<15m

[0128]

[0129] In the formula, ΔL is the increase in the length of the suspension rod under constant load. L1=L+ΔL.

[0130] Step fourteen: Based on the obtained and processed data, establish a bridge deck mechanical model. During this process, measure the inclination angle between the bridge deck and the cables, the edge distance, the center distance, and the cable base spacing. Calculate relevant mechanical data (axial force, bending moment, etc.) for the bridge structure. Divide the bridge into multiple regions based on the influence of each cable on the tensile force. Then, perform mechanical analysis and calculations on each micro-element region. Using the relationship between internal and external forces in mechanics of materials, derive the internal forces of the bridge in each region. Integral calculations are then used to obtain the internal forces at various locations throughout the entire bridge. Specifically, calculate the relevant mechanical data (axial force, bending moment, etc.) for the entire bridge. Assume the cable inclination angle is θ, the edge distance is l1, the center distance is l2, and the cable base spacing is a. All cables are under tension. Recorded as negative, F NK Let be the axial force from the k-th cable to the (k+1)-th cable, then:

[0131]

[0132] Will Let M(x) be positive. k This represents the bending moment at a distance x from the origin in the k-th segment. The first segment consists of the section from the first cable to the second cable, and so on. In particular, the k-th segment... The root line reaches point B as the first Section, from point B to The root is the first The first segment, and so on, with the middle segment drawn separately.

[0133]

[0134] Similarly, the bending moment in the lower half can be obtained. The bending moment in the middle half is:

[0135]

[0136] According to the formula Get |σ max |=σ tmax ≤[σ t ];|σ min |=σ cmax ≤[σ c ].

[0137] Step 15: Based on the obtained position-chromaticity vector (x, y, R, G, B) of the detection points and the calculated relationship between the bridge cable frequency and cable force, the relationship between the bridge deck force and internal force will be compared with the corresponding data in the initial database to preliminarily determine whether the bridge has defects.

[0138] Step sixteen: In the analysis of the location-chromaticity vector, the associated spatiotemporal coordinate information is embedded into the physical structure of the image and color to achieve the fusion of associated multi-data elements and form a complete physical model of the bridge. The chromaticity distribution of the bridge layer obtained after denoising the bridge image constitutes an RGB chromaticity sample set. Then, the chromaticity sample set is deduplicated, and the chromaticity junctions are marked and integrated into a comparison image library. The measured real-time location-chromaticity vector is compared with the chromaticity of the same location in the original database. Different colors and texture elements are marked in the physical model of the bridge. At the same time, relevant captured images from the mobile app are extracted and the real-scene image of the area is displayed. The operator determines the location of defects on the bridge appearance based on the real-scene image. If there is no significant difference between the chromaticity and the data in the original database, the bridge appearance is free of defects.

[0139] Step 17: Represent the cable force, vibration frequency, axial force, bending moment, and other mechanical elements generated by the bridge under external forces in the bridge physical model from the bridge structural data. Compare this with the safe vibration frequency and cable force, maximum tensile strength, and maximum bending strength of the bridge in the initial database. Simultaneously, extract relevant images from the mobile app and display the real-world image of the area. Operators can determine the location of defects in the bridge's internal structure based on the real-world image. If the vibration frequency of the cable differs significantly from its natural frequency, or if the stress caused by the external forces on the bridge exceeds the ultimate strength, it indicates that the bridge has been damaged or is at risk of damage at the corresponding location, and a damage warning can be issued.

[0140] Step 18: Compare the test set with the initial dataset. Establish a chromaticity difference threshold K based on historical disease data. When the difference between the chromaticity vector (x,y,R,G,B) of a monitoring point and the initial chromaticity vector (x,y,R,G,B) in the initial database exceeds the threshold K, the monitoring point is considered a disease location. When the difference does not exceed the threshold K but exceeds the 90% guaranteed value (i.e., the difference is greater than 0.9K), the point is considered a potential disease location and requires manual inspection. The detailed flowchart for chromaticity identification and disease location is as follows: Figure 7 As shown.

[0141] Step 19: In establishing the surface defects and internal damage of the bridge, the structural parameters of the bridge, namely cable tension, vibration frequency, sag, bridge deck displacement, vibration frequency, chromaticity change, and cable displacement, are assigned weights of W1, W2, W3, W4, W5, W6, and W7, respectively. The weight settings for the bridge structural indicators are shown in Table 1.

[0142]

[0143] Table 1

[0144] Step 20: Evaluate the data by comparing the collected and processed structure and color data with the information in the database. The evaluation method mainly involves determining whether parameter i exceeds a threshold using a 0 / 1 judgment. If indicator i is within the safety limit... Judgment value B of indicator i i A value of 1 indicates the indicator is in a normal state; indicator i exceeds the 90% safety limit. Judgment value B of indicator i i A value of 0 indicates that the corresponding structural parameter is in an abnormal state. Simultaneously, a weighted average score for the bridge structure is calculated based on the parameter weights to determine the structural damage level. The structural hazard level determination is shown in Table 2, and the framework for indicator assessment and control procedures is as follows: Figure 8 As shown, the structural score is calculated as shown in formula (4).

[0145]

[0146] The judgment value matrix after indicator 0 / 1 is represented as follows:

[0147]

[0148] Among them, G t Represents the structural score, A i The actual measured value of the representative indicator. B represents the 90% safety expectation limit of the indicator design. i W represents the judgment value of each indicator. i The values ​​represent the weights of each indicator, q = 1, 2, 3, 4, 5, 6. These represent the cable vibration frequency, cable force, bridge deck displacement, bridge deck vibration frequency, chromaticity change, and cable displacement, respectively. The structural indicator weights are obtained according to bridge inspection and monitoring standards, and the specific values ​​are shown in Table 1. The structural score is located in the range [0, 100], with a higher score indicating a safer overall bridge structure.

[0149] Step 21: This monitoring method uses a seven-day monitoring cycle, and measures, processes, and statistically analyzes the data within the measurement cycle. On day i, three predictions are made, and structural scores are calculated for each of the three measurements, resulting in a vector containing the structural scores from the three measurements for day i. Assumption i = 1, 2, 3…7. We choose the structural score representative value G for day i. ti As the true score value of the structure on day i, the representative score value G of the structure on day i is... ti The calculation process is as follows:

[0150] 1. and respectively with Perform the difference. when When the condition is constant, select the cubic structural score. The arithmetic mean of the values ​​is used as the representative structural score G for day i. ti .

[0151] 2. When there exists a set j such that The other set of j satisfies When, take the median value. G, representing the structural score on day i. ti .

[0152] 3. When If the condition is not met, the vector of the measurement structure score should not be used as the basis for evaluation, and three more measurements need to be performed.

[0153] Based on the above method, seven representative structural scores [G] can be obtained within a monitoring period. t1 G t2 G t3 G t4 G t5 G t6 G t7 ].

[0154] Step 22: Since indicators such as cable vibration frequency, cable sag, and bridge deck displacement are related to time variations, a predictive model can be established using time series analysis on the data from the eighth day outside the monitoring period. Let the original time series data be denoted as G. ti (i = 1, 2, ..., 7), to prevent spurious regression problems due to non-stationary data, we first test the stationarity of the data. For a given time series sample G... ti (i=1,2,…,7), denoted by G ti The rank of R is t =R(G ti Consider the variable pair (t, R). t The Spearman correlation coefficient q for t = 1, 2, ..., n s ,have:

[0155]

[0156] Constructing statistics:

[0157]

[0158] If the time series is non-stationary, in order to construct a stationary series, we can adjust the time series G. ti Perform a first-order difference operation b on (i = 1, 2, ..., 7) t =a t+1 -a t A stable sequence b is obtained. t .

[0159] Finally, the stable sequence is decomposed into long-term trend variation, cyclical variation, and irregular variation. A product algorithm is used, setting the final indicator variation as: Long-term trend variation × Cyclical variation × Irregular variation. A regression-moving average algorithm is then employed, combining the autoregressive and moving average processes to generate an iterative equation, yielding the predicted indicator for the eighth day, i.e., the structural score representative value G for the eighth day. t8 .

[0160] Step 23: Using the method described above, obtain a set of predicted data including the monitoring period and the second day of the monitoring period, resulting in eight representative structural score values. Monitoring indicators A and B can be obtained from these eight representative structural score values. The calculation methods for monitoring indicators A and B are as follows: Filter the eight sets of data and calculate the representative structural score value G. ti The percentage of scores greater than 90 is denoted as monitoring indicator A; the representative value G of the structured score is calculated. ti The percentage of scores below 80 is designated as monitoring indicator B. The higher the hazard level, the more dangerous the structure. The relationship between monitoring indicators and structural hazard levels is shown in Table 2.

[0161] Monitoring indicators Hazard level A ≤ 0.12 and B ≥ 0.75 Ⅴ A≤0.38 and B≥0.38 Ⅳ Other situations Ⅲ A ≥ 0.38 and B ≤ 0.38 Ⅱ A ≥ 0.75 and B ≤ 0.12 Ⅰ

[0162] Table 2

[0163] Step 24: The bridge maintenance plan and traffic control plan will be implemented through a tiered control approach. Specific steps for structural condition assessment and maintenance control will be outlined below. Figure 8 Based on the structural hazard level and structural parameter scores, bridge maintenance and traffic control plans are generated, and adjustments are implemented according to the plans, as shown in Tables 3 and 4. Appropriate maintenance and traffic control plans are selected based on the structural hazard level. When the hazard level is I, the bridge structure is in a safe, stable, and defect-free state, requiring no special monitoring or maintenance, and vehicles are allowed to pass normally. When the bridge hazard level is II, close monitoring and traffic flow control are required. Maintenance and traffic control plans for hazard levels I and II are shown in Table 3. When the hazard levels are III, IV, and V, the overall score alone is insufficient to determine the hazard location and maintenance plan; structural index parameters are used to determine the maintenance and traffic control plans. Specific maintenance and traffic control plans are shown in Table 4.

[0164]

[0165] Table 3

[0166]

[0167] Table 4

[0168] Step 25: After providing the bridge maintenance plan, monitor and record the changes in various structural indicators in real time after the plan's implementation. Using closed-loop control theory, compare the structural changes after the measures are implemented with the expected results, and adjust the weighting coefficients of the corresponding structural parameters based on the comparison results. When the next structural damage or defect warning occurs, select a reasonable bridge maintenance and control plan based on the structural score, and adjust the plan's control measures according to the new weighting coefficients. After each plan implementation, it is necessary to update the weighting coefficients of the adjusted measures promptly based on the output feedback to improve the stability of the calculation model and the accuracy of the measure adjustments.

[0169] Step 26: For the monitoring data on bridge safety, the physical indicators such as temperature, displacement, and strain obtained through sensors are processed and transmitted to the application's cloud server. From there, the data is distributed to other users' mobile apps, ensuring secure information transmission and facilitating staff monitoring the bridge's safety status. The application allows bridge maintenance personnel to easily monitor the bridge, and its security must be guaranteed. This application includes a login interface for staff to access the application using a username and password. If either is entered incorrectly, a message "Username or password incorrect" will be displayed; successful login occurs when both are correct. After successful login, staff can also check in on their mobile phones. A push notification mechanism is added to the app to send information to staff when bridge safety issues arise. Simultaneously, some bridge-related information can be shared on platforms such as QQ and WeChat, allowing more people to learn about bridge-related information through relevant public accounts or push notifications.

[0170] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

[0171] A computer-readable storage medium storing one or more programs, the programs including instructions that, when executed by a computing device, cause the computing device to perform a method for monitoring bridge defects and structural damage based on a mobile app and machine vision.

[0172] A computing device includes one or more processors, one or more memories, and one or more programs, wherein the one or more programs are stored in the one or more memories and configured to be executed by the one or more processors, and the one or more programs include instructions for performing a method for monitoring bridge defects and structural damage based on a mobile app and machine vision.

[0173] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0174] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0175] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0176] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0177] The above are merely embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of the claims of the present invention pending approval.

Claims

1. A method for monitoring bridge surface defects and structural damage based on mobile app and machine vision, characterized in that: Using mobile devices to capture video images of bridges, the appearance and structural parameters of the bridges are extracted from the video images; Establish a three-dimensional spatial coordinate system on the bridge deck, obtain the coordinates of each position of the bridge, and merge the bridge position coordinates with the bridge appearance parameters to obtain the appearance parameters corresponding to each position. The appearance parameters corresponding to each location are compared with the initial appearance parameters of the bridge. If the difference between the comparison at that location exceeds the threshold, it indicates that there is a defect at that location, and the location of the defect is displayed on the mobile app. A bridge mechanical model is constructed based on the bridge's structural parameters. The mechanical information of the bridge is obtained through the mechanical model. If the mechanical information of the bridge exceeds the safety limit of the bridge itself, it indicates that there is damage to the bridge structure. Input the bridge's appearance parameters, structural parameters, and mechanical information into a pre-built structural scoring model, and output the bridge's safety rating. The appearance parameters of a bridge include RGB chromaticity and texture feature lines; the structural parameters of a bridge include the displacement and vibration frequency of the bridge deck and the displacement and vibration frequency of the cables. The scoring method for the structural scoring model is as follows: The indicator i is determined by whether it exceeds the threshold (0 / 1), indicating that the indicator i is within the safety limit. Judgment value B of indicator i i A value of 1 indicates the indicator is in a normal state; indicator i exceeds the 90% safety limit. Judgment value B of indicator i i A value of 0 indicates that the corresponding structural parameter is in an abnormal state. At the same time, the weighted average score of the bridge structure is calculated based on the parameter weights to determine the damage level of the structure; The structural score is calculated as shown in formula (4). The judgment value matrix after indicator 0 / 1 is represented as follows: Among them, G t Represents the structural score, A i The actual measured value of the representative indicator. B represents the 90% safety expectation limit of the indicator design. i W represents the judgment value of each indicator. i The values ​​represent the weights of each indicator, q = 1, 2, 3, 4, 5, 6; representing cable vibration frequency, cable force, bridge deck displacement, bridge deck vibration frequency, color change, and cable displacement, respectively. The weights of the structural indicators are obtained according to bridge inspection and monitoring standards. The structural score is located in the range of [0, 100]. The higher the score, the safer the bridge structure is overall.

2. The method for monitoring bridge surface defects and structural damage based on mobile app and machine vision according to claim 1, characterized in that: The process of extracting RGB chromaticity and texture feature lines is as follows: The video frame segmentation algorithm in OpenCV is called, and the video path to be segmented and the file path to save the images are passed in as two parameters. The vibration information of each frame of the video file is obtained from the video file and the video keyframes are extracted. Based on histogram differences, the Canny operator for edge detection is used to obtain the shape features of the bridge deck and cables, non-interest regions are clipped, RGB color information of the image is extracted, and feature lines are drawn. Finally, based on the 3σ rule, the probability that the pixel chromaticity distribution is in (μ-1.5σ,μ+1.5σ) is 0.9330. The threshold is set at the 93.3% chromaticity quantile of all pixels, and each frame of the image is binarized, that is, the chromaticity is adjusted to 0 or 255.

3. The method for monitoring bridge surface defects and structural damage based on mobile app and machine vision according to claim 2, characterized in that: 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 local maxima of the gradient magnitude are detected. Then, a low threshold T1 is used to obtain weak edges E1, and similarly, a high threshold T2 is used to obtain edges E2. Only connected components that are connected to E2 are retained as output edges E in E1.

4. The method for monitoring bridge surface defects and structural damage based on mobile app and machine vision according to claim 2, characterized in that: The displacement extraction process for the bridge deck and cables is as follows: During video framing, an image stabilization algorithm based on mobile phone accelerometer data extraction is used. By capturing and identifying fixed points in the image, the relative displacements of the bridge deck, cables, and fixed points are obtained, yielding relative vibration data. This data is then used to correct the image displacement and reduce displacement errors caused by mobile device shaking. The displacement data of the bridge and cables is subtracted from the data of the fixed points to obtain the actual vibrations of the bridge and cables relative to the ground. Finally, the acceleration data from the Orion-CC application is extracted, based on T=4ml. 2 (Δf) 2 , and obtain the corrected relative displacement data.

5. The method for monitoring bridge surface defects and structural damage based on mobile app and machine vision according to claim 1, characterized in that: The noise reduction process for the displacement and vibration frequency of the bridge deck and the displacement and vibration frequency of the cables is as follows: Wavelet denoising is used to perform wavelet transform on the noise points in the data to obtain the high-frequency values ​​and approximate low-frequency values ​​of each noise point. Thresholding is performed on the high-frequency values, and wavelet reconstruction is performed on each frequency value component after processing to obtain the denoised image. Let the noisy image model a(i,j) be: a(i,j)=b(i,j)+c(i,j) In the formula, a(i,j) is the noisy image after it has been corrupted by noise, b(i,j) is the initial noise-free image, and c(i,j) is Gaussian noise that follows a Gaussian normal distribution. After performing wavelet transform on both sides of the equation, we obtain the coefficients after wavelet transform: IN a =In b +W c W a W represents the coefficients of the original image after wavelet transform. b W represents the coefficients of the original noise-free image after wavelet transform. c The coefficients are the Gaussian white noise after forearm transformation; IN b =In c -IN a In wavelet transform-based thresholding denoising methods, the threshold function can be selected from the following two options: The hard threshold function is shown in equation (1), W a These are the wavelet coefficients after thresholding, where λ is the threshold value. When the absolute value of a wavelet coefficient is less than the given threshold, the value is set to 0; when the wavelet coefficient is greater than the threshold, the original value is retained, resulting in the denoised wavelet coefficients. The soft thresholding function, as shown in formula (2), sets the value to 0 when the absolute value of the wavelet coefficient is less than the given threshold, and subtracts the threshold from the value of the wavelet coefficient when the wavelet coefficient is greater than the threshold, thus obtaining the denoised wavelet coefficients. An adaptive threshold selection method is adopted, which dynamically adjusts the threshold and output value according to different wavelet coefficients, as shown in formula (3): P is a regulating factor. This represents the wavelet coefficients after adaptive thresholding, where T is the threshold set by the function.

6. The method for monitoring bridge surface defects and structural damage based on mobile app and machine vision according to claim 1, characterized in that: The bridge mechanics model includes a bridge cable mechanics model and a bridge deck mechanics model.

7. The method for monitoring bridge surface defects and structural damage based on mobile app and machine vision according to claim 6, characterized in that: The process of constructing the mechanical model of bridge cables is as follows: When the bridge boundary constraints are simply supported, considering the effects of bending stiffness and the variation in suspension bridge length, and based on actual calculations of the cable forces in a certain tied-arch bridge with under-deck construction, the differential equation for cable vibration is derived using string vibration theory without considering the effect of sag. Where EI is the bending stiffness, T is the cable force, m is the mass per unit length of the cable, and y(x,t) is the displacement function of each point on the cable as a function of time. Solving this equation yields... Y(X)=Asin(αx)+Bcos(αx)+Csinh(βx)+Dcosh(βx) Where A, B, C, and D are constant terms, ω is the natural frequency, and α = [(p 4 +q 4 ) 0.5 -p 2 ] 0.5 ;β=[(p 4 +q 4 ) 0.5 +p 2 ] 0.5 ;p 2 =T / (2EI); q 4 =ρω 2 / (2EI); The following formula can be obtained: Formula considering the influence of bending stiffness: T=4mL 2 f 2 (n) / n 2 -EI(nΠ / L) 2 Formula without considering the effect of bending stiffness: T=4ρmL 2 f 2 (n) / n 2 In the formula, T is the cable tension; m is the cable density; L is the calculated length of the cable; f(n) is the nth natural frequency of the cable; and n is the order of the natural frequency. Taking into account the influence of bending stiffness and length increase, a correction coefficient θ is introduced to modify the calculation formula, resulting in a cable force calculation formula suitable for short suspenders. The fitting function formula is: T=θ×4mL 2 f 2 (n) / n 2 By solving the equation and introducing correction coefficients that consider the changes in bending stiffness and bridge length, the cable force, natural frequency, and calculated length of the completed bridge are substituted into the above formula for regression calculation. The calculated value is θ = 0.

962. Therefore, the corrected formula for calculating the short suspension rod is: T=3.848mL 2 f 2 (n) / n 2 ; The formula for calculating cable force, taking into account both bending stiffness and the effect of the increase in suspender length, is as follows: T=3.848mL 2 f 2 (n) / n 2 ,L<15m In the formula, △L is the increase in the length of the hanger under constant load, L1=L+△L.

8. The method for monitoring bridge surface defects and structural damage based on mobile app and machine vision according to claim 6, characterized in that: The process of constructing the bridge deck mechanical model is as follows: Calculate the relevant mechanical data for the entire bridge. Assume the cable inclination angle is θ, the edge distance is l1, the middle distance is l2, and the spacing between the cable bases is a. All cables are under tension. Recorded as negative, F NK Let be the axial force from the k-th cable to the (k+1)-th cable, then: Will Let M(x) be positive. k Let x represent the bending moment at a distance x from the origin in the k-th segment. The segment from the first cable to the second cable is the first segment, and so on. The root line reaches point B as the first Section, from point B to The root is the first The first segment, and so on, with the middle segment drawn separately; in Similarly, the bending moment of the lower half is obtained, and the bending moment of the middle half is: According to the formula get max |=s tmax ≤[σ t ];|s min |=s cmax ≤[σ c ]。

Images