Cable-stayed bridge cable force intelligent monitoring method and system based on multi-modal data fusion
By using a multimodal data fusion method, combining grayscale images and strain data, the linear deviation, stress nonuniformity, nonlinear strength, and energy mobility of the cable-stayed bridge are calculated. This solves the problem of inaccurate assessment in traditional cable-stayed bridge cable force monitoring and enables accurate assessment and safety assurance of the cable force state of cable-stayed bridges.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ROAD & BRIDGE INT CO LTD
- Filing Date
- 2026-05-08
- Publication Date
- 2026-06-05
AI Technical Summary
Traditional methods are insufficient to fully reflect the spatial mechanical state of the entire cable in cable-stayed bridge cable force monitoring. Single-dimensional strain data lacks the ability to capture cable deformation, resulting in a single dimension for judging cable force anomalies. This can easily lead to monitoring misjudgments and response delays, affecting the accuracy and reliability of the assessment.
A multimodal data fusion method is adopted to acquire grayscale images and strain data of each cable of a cable-stayed bridge, calculate the alignment deviation, stress non-uniformity, nonlinear strength and energy mobility, and evaluate the cable stress state of the cable by combining the anomaly coefficient, so as to achieve multi-dimensional monitoring.
Accurately capturing abnormal changes in cable force improves the precision of cable force monitoring for cable-stayed bridges, ensuring their safe operation. It can also identify structural damage, material aging, and local defects, reducing monitoring errors.
Smart Images
Figure CN122153745A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of bridge structural health technology, specifically to a smart monitoring method and system for cable-stayed bridge cable forces based on multimodal data fusion. Background Technology
[0002] As a major form of modern long-span bridge, cable-stayed bridges rely heavily on cable tension as a key parameter for assessing the structural health and safety of the bridge. Abnormal changes in cable tension can directly reflect structural damage, material degradation, or sudden changes in external loads. During long-term service, cable-stayed bridge cables inevitably face various complex factors such as corrosion, fatigue, and vibration, leading to gradual performance degradation. Therefore, long-term and precise monitoring of cable tension is necessary to ensure the safe operation of cable-stayed bridges.
[0003] Traditional methods for monitoring cable forces in cable-stayed bridges are insufficient to comprehensively reflect the overall spatial mechanical state of the entire cable. They are insensitive to changes in the overall cable alignment, local stress concentrations, and damage in non-measurement areas. Furthermore, single-dimensional strain data lacks the ability to capture spatial morphology such as cable deformation, resulting in a limited dimension for judging cable force anomalies. When dealing with sudden loads or gradual damage propagation, this can easily lead to monitoring misjudgments and delayed responses, affecting the accuracy and reliability of cable force assessment for cable-stayed bridge structures. Summary of the Invention
[0004] To address the aforementioned technical challenges, a smart monitoring method and system for cable-stayed bridge cable forces based on multimodal data fusion is provided to resolve existing issues.
[0005] The solution to the technical problem presented in this application is to provide a smart monitoring method and system for cable-stayed bridge cable forces based on multimodal data fusion, comprising the following steps: In a first aspect, embodiments of this application provide a smart monitoring method for cable-stayed bridge cable forces based on multimodal data fusion, the method comprising the following steps: Acquire grayscale images and strain data of each stay cable on the cable-stayed bridge at each time point within each monitoring cycle; Extract the cable skeleton from the grayscale image; for each cable in each monitoring period, analyze the differences in the bending features of pixels on the cable skeleton in the grayscale image at different times, as well as the differences in the distribution of the overall spatial shape of the cable skeleton, and calculate the linear deviation of each cable in each monitoring period. Based on the fluctuation of gray values of pixels on the cable frame in grayscale images at different times, and the inconsistency of gradient direction distribution of pixels on the cable frame, the stress non-uniformity of each cable in each monitoring period is calculated. Combined with the linear deviation, the first evaluation value of each cable in each monitoring period is obtained. The distribution dispersion of strain data peaks and their long-term trend, as well as the deviation of strain data from linear trends, are evaluated for each cable in each monitoring period. The nonlinear intensity of each cable in each monitoring period is calculated. By analyzing the energy difference in the frequency domain distribution of strain data within the local area where different peaks are located, the energy mobility of each cable in each monitoring period is calculated. Combined with the nonlinear intensity, the second evaluation value of each cable in each monitoring period is determined. Based on the first and second evaluation values, the anomaly coefficient of each cable is determined for each monitoring period, and the anomaly assessment of the cable force state of the cable-stayed bridge is carried out.
[0006] Preferably, the calculation of the alignment deviation of each stay cable in each monitoring period includes: For each cable in each monitoring period, curve fitting is performed on the position coordinates of all pixels on the cable skeleton in the grayscale image at each time point to obtain the fitted curve, and the mean value of the curvature of all pixels on the fitted curve is calculated as the average curvature; the sum of the differences in the average curvature of the grayscale images at any two times in each monitoring period is calculated. Calculate the sum of the correlation between the grayscale images and the fitted curves for any two times within each monitoring period, and then perform a positive mapping on them; The linear deviation is the ratio of the sum to the result of the positive mapping.
[0007] Preferably, the calculation of the stress non-uniformity of each stay cable in each monitoring cycle includes: For each cable in each monitoring period, calculate the dispersion of gray values of all pixels on the cable skeleton in the grayscale image at each time point; calculate the average value of the dispersion of grayscale images at all times in each monitoring period. Obtain the gradient directions of all pixels on the cable-stayed bridge frame and form a gradient sequence; calculate the sum of the JS divergence of the gradient sequences corresponding to any two grayscale images within each monitoring period, and denote it as the gradient difference; The stress nonuniformity is the product of the average value and the gradient difference.
[0008] Preferably, the first evaluation value is the product of the linear deviation and the stress non-uniformity.
[0009] Preferably, the calculation of the nonlinear strength of each stay cable in each monitoring period includes: Obtain the peak values of strain data for each cable at all times during each monitoring period; calculate the coefficient of variation and Hurst exponent of the peak values of all peak values for each cable during each monitoring period. Linear fitting was performed on the strain data of each cable at all times during each monitoring period, and the fitting error was calculated. The nonlinear intensity is the product of the fitting error, the coefficient of variation, and the Hurst exponent, which are positively mapped together.
[0010] Preferably, the calculation of the energy mobility of each stay cable in each monitoring cycle includes: For each cable, under each monitoring cycle, multiple moments in the neighborhood of each wave peak are recorded as local time periods; frequency domain analysis is performed on the strain data of all moments within the local time period to obtain the spectrum corresponding to each wave peak, and the energy of all frequency components in the spectrum is used to form an energy sequence. The energy mobility is the sum of the differences in the energy sequences of any two peaks for each cable in each monitoring period.
[0011] Preferably, the second evaluation value is the product of nonlinear intensity and energy mobility.
[0012] Preferably, determining the anomaly coefficient of each cable in each monitoring period includes: using a comprehensive evaluation algorithm to comprehensively evaluate the first evaluation value and the second evaluation value of each cable in each monitoring period, and obtaining the comprehensive score of each cable in each monitoring period as the anomaly coefficient of each cable in each monitoring period.
[0013] Preferably, the abnormal assessment of the cable force state of the cable-stayed bridge includes: obtaining the segmentation threshold of the abnormal coefficient of all the cable-stayed bridge cables in each monitoring period; recording the cable-stayed cables with an abnormal coefficient greater than the segmentation threshold as abnormal cable-stayed cables; if each cable is an abnormal cable-stayed cable in multiple consecutive monitoring periods, then the cable force state of the cable-stayed cable is abnormal, otherwise, the cable force state of the cable-stayed cable is not abnormal.
[0014] Secondly, embodiments of this application also provide a smart monitoring system for cable-stayed bridge cable forces based on multimodal data fusion, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any of the above-described smart monitoring methods for cable-stayed bridge cable forces based on multimodal data fusion.
[0015] This application has at least the following beneficial effects: This application monitors and evaluates the cable force of stay cables from both spatial and temporal dimensions by acquiring images and strain data of the stay cables at different times, avoiding the inaccuracy of monitoring the cable force using single data. It calculates the linear deviation of each stay cable in each monitoring period. The beneficial effect is that it considers the bending of pixels on the cable skeleton and the differences in the overall spatial shape at different times, reflecting the shape changes of the stay cables. This allows for accurate capture of abnormal changes in the macroscopic geometry of the cables caused by cable force imbalance, assessing the possibility of significant external force influence or structural problems. It also calculates the stress unevenness of each stay cable in each monitoring period. Uniformity, its beneficial effect lies in considering the grayscale fluctuations and gradient direction inconsistencies of pixels on the cable skeleton, which can effectively identify surface texture variations and edge irregularities caused by stress concentration or local material aging; obtaining the first evaluation value of each cable in each monitoring period, its beneficial effect lies in linking the two physical dimensions of linear shape and surface condition, comprehensively reflecting the spatial deformation characteristics of the cable, and preliminarily assessing the possibility of abnormal mechanical states of the cable, reflecting possible structural damage, material aging, or local defects; calculating the nonlinear strength of each cable in each monitoring period, its beneficial effect lies in considering The nonlinear characteristics of strain data, along with the discrete distribution and long-term trend of peaks, reflect the severity of load abrupt changes or whether the damage has a continuous deterioration inertia. This assesses the severity of sudden loads or damage propagation in the stay cables, indicating that the cable stress is in an unstable and continuously deteriorating dangerous state. Calculating the energy mobility of each stay cable in each monitoring period is beneficial because it allows for analysis of the differences in energy distribution across different strain abrupt events in the frequency domain, determining the consistency of the nature of abnormal events, reflecting the significant changes in stress distribution in the stay cables, and identifying localized stress inhomogeneity or material aging phenomena. It also helps determine the energy mobility of each stay cable in each monitoring period. The second evaluation value of the cable-stayed bridge has the advantage of combining the trend characteristics in the time domain with the response characteristics in the frequency domain, which has a strong ability to capture cable force anomalies caused by sudden loads or internal damage, thus compensating for the shortcomings of image monitoring in terms of time resolution. It determines the anomaly coefficient of each cable-stayed bridge in each monitoring period and evaluates the anomaly of the cable force state of the cable-stayed bridge. Its advantage lies in the final fusion of the first evaluation value, which represents spatial deformation, and the second evaluation value, which represents temporal dynamics, thus realizing the complementarity and verification of multi-source information at a higher level, improving the accuracy of monitoring cable force anomalies in cable-stayed bridges, and effectively ensuring the safe operation of cable-stayed bridges. Attached Figure Description
[0016] The following section provides a more detailed description of the intelligent monitoring method for cable-stayed bridge cable forces based on multimodal data fusion, in conjunction with the accompanying drawings.
[0017] Figure 1A flowchart illustrating the steps of the intelligent monitoring method for cable-stayed bridge cable force based on multimodal data fusion provided in this application embodiment; Figure 2 A flowchart illustrating the steps of the method for obtaining anomaly coefficients provided in this application embodiment. Detailed Implementation
[0018] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description of the intelligent monitoring method and system for cable-stayed bridge cable forces based on multimodal data fusion, in conjunction with the accompanying drawings and implementation examples, is provided. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.
[0020] Please see Figure 1 The diagram illustrates a flowchart of a smart monitoring method for cable-stayed bridge cable forces based on multimodal data fusion, according to an embodiment of this application. The method includes the following steps: Step 1: Obtain grayscale images and strain data of each stay cable on the cable-stayed bridge at each time point within each monitoring cycle.
[0021] During long-term service, cable-stayed bridges are subjected to various variable loads, such as traffic loads, wave loads, and wind loads. Under these variable loads, the stress on the stay cables changes in real time. Therefore, monitoring the time-varying cable force on the stay cables of cable-stayed bridges is the basis for assessing the structural condition and fatigue damage of the stay cables.
[0022] By deploying multiple high-definition industrial cameras at the top of the bridge towers or piers of the cable-stayed bridge, each camera can cover the entire connection area between the cable head and the cable body of a cable, and acquire grayscale images of each cable in real time; and by deploying fiber optic strain sensors on each cable of the cable-stayed bridge, strain data can be collected in real time. Multiple moments are treated as a monitoring cycle, and grayscale images and strain data of each cable are obtained at different moments within each monitoring cycle. In this embodiment, the monitoring period is 30 minutes. Two grayscale images are acquired every minute. The acquisition frequency of the fiber optic strain sensor is 100 Hz. As for other implementation methods, the implementer can set them according to the actual situation.
[0023] The Laplacian filtering algorithm is used to enhance the acquired grayscale image, and the Zhang calibration method is used to correct the distortion and pose of the grayscale image. The Laplacian filtering algorithm and the Zhang calibration method are well-known techniques and will not be described in detail here.
[0024] After filling missing values in the strain data, normalization is performed, and the grayscale image is time-aligned with the strain data. In this embodiment, cubic spline interpolation is used to handle missing values, and the maximum-minimum method is used for normalization. The cubic spline interpolation, maximum-minimum method, and time alignment process are all well-known techniques and will not be described in detail here.
[0025] Thus, grayscale images and strain data of each stay cable on the cable-stayed bridge at different times during each monitoring cycle were obtained.
[0026] Step 2: Extract the cable skeleton from the grayscale image; for each cable in each monitoring period, analyze the differences in the bending characteristics of pixels on the cable skeleton at different times in the grayscale image, as well as the differences in the distribution of the overall spatial shape of the cable skeleton, and calculate the line deviation of each cable in each monitoring period; based on the fluctuation of grayscale values of pixels on the cable skeleton at different times in the grayscale image, and the inconsistency in the distribution of gradient directions of pixels on the cable skeleton, calculate the stress non-uniformity of each cable in each monitoring period, and combine it with the line deviation to obtain the first evaluation value of each cable in each monitoring period.
[0027] When monitoring the cable force of a cable-stayed bridge, the more severe the abnormality in the cable force, the greater the degree of deviation in the cable shape caused by abnormal cable deformation. In other words, the more significant the positional change of the cable within the monitoring period, and the greater the difference in the bending state of the cable at different times. At the same time, due to the uneven stress on the cable or the local aging of the cable material, the gray values of the pixels represented by the cable in the grayscale image show significant fluctuations, and the gradient direction inconsistency of the pixels on the corresponding skeleton of the cable in the image is also higher.
[0028] Based on the above analysis, the positional changes of the cable-stayed bridge in the grayscale images at different times within each monitoring period are analyzed, and the linear deviation is calculated, specifically: A semantic segmentation algorithm is used to obtain the cable-stayed bridge region in the grayscale image; after edge detection of the cable-stayed bridge region, skeleton extraction is performed to obtain the cable-stayed bridge skeleton in the grayscale image; In this embodiment, the Mask R-CNN semantic segmentation algorithm is a well-known technology and will not be described in detail here. Secondly, the Canny edge detection algorithm is used for edge detection, and the Zhang-Suen skeleton extraction algorithm is used for skeleton extraction. Both the Canny edge detection algorithm and the Zhang-Suen skeleton extraction algorithm are well-known technologies and will not be described in detail here.
[0029] For each cable in each monitoring period, curve fitting is performed on the position coordinates of all pixels on the cable skeleton in the grayscale image at each time point to obtain the fitted curve, and the mean value of the curvature of all pixels on the fitted curve is calculated as the average curvature. In this embodiment, the least squares method is used for curve fitting. The least squares method and the calculation of curvature are well-known techniques and will not be described in detail here.
[0030] For each cable, calculate the sum of the correlation between the grayscale images and the fitted curves at any two times within each monitoring period, and then perform a positive mapping on them; In this embodiment, the correlation is calculated by taking the cosine similarity of the fitted curves corresponding to the grayscale images of each cable at any two times within each monitoring period. The calculation of cosine similarity is a well-known technique and will not be elaborated upon here. Secondly, the positive mapping process is as follows: an exponential function is used for positive mapping, assuming the sum of the correlation is denoted as... ,but The result is taken as the result of the positive mapping, where, Let be an exponential function with the natural constant as its base. Through the process of positive mapping, the result of the positive mapping is made to be greater than 0.
[0031] Calculate the sum of the differences in the average curvature of the grayscale images at any two times within each monitoring period; In this embodiment, the sum of the absolute values of the differences in the average curvature of the grayscale images at any two times within each monitoring period is calculated.
[0032] The ratio of the summation to the positive mapping result is used as the linear deviation of each cable in each monitoring period. It should be noted that the larger the average curvature, the more significant the bending of the cable in the grayscale image at that moment; the larger the sum, the greater the change in the cable shape at different times, which may be affected by a large external force, resulting in obvious deformation; the smaller the positive mapping result, the more inconsistent the overall spatial shape and orientation of the cable throughout the monitoring period, i.e., the line shape has shifted; the greater the obtained line shape deviation, the larger the shape change of the cable, and the lower the consistency of this change between different times, which may be affected by a large external force or have structural problems.
[0033] Secondly, the fluctuation of grayscale values in the neighborhood of pixels on the cable-stayed bridge frame in the grayscale image, as well as the change of gradient direction of pixels on the cable-stayed bridge frame, are analyzed to calculate stress non-uniformity. Specifically: For each cable in each monitoring period, calculate the dispersion of gray values of all pixels on the cable skeleton in the grayscale image at each time point; calculate the average value of the dispersion of grayscale images at all times in each monitoring period. In this embodiment, the degree of dispersion is measured by calculating the variance of the gray values of all pixels on the cable-stayed bridge skeleton. As other implementation methods, implementers may use other methods of the prior art, such as standard deviation, etc. This embodiment does not impose any special restrictions on this.
[0034] Obtain the gradient directions of all pixels on the cable-stayed bridge skeleton and form a gradient sequence; It should be noted that the gradient direction is a well-known technique and will not be elaborated upon here.
[0035] Calculate the sum of the JS divergence of the gradient sequences corresponding to all two grayscale images within each monitoring period, and denote it as the gradient difference; It should be noted that the calculation of JS divergence is a well-known technique, and will not be elaborated upon here.
[0036] The product of the average value of each cable in each monitoring period and the gradient difference is used as the stress non-uniformity of each cable in each monitoring period. It should be noted that the greater the degree of dispersion, the larger the average value, indicating that the grayscale value of the pixels on the cable frame in the grayscale image at that time has a large fluctuation, indicating that the difference in grayscale value of the pixels on the cable frame is higher, reflecting that the overall stress distribution of the cable is uneven or there is local damage. The greater the gradient difference, the greater the difference in the gradient direction of the pixels on the cable frame in the grayscale image at different times, reflecting that the stress distribution of the cable is significantly different, and there may be local stress unevenness or material aging. The greater the obtained stress unevenness, the more the surface brightness distribution of the cable is constantly changing, and the edge geometry of the cable is also constantly changing, reflecting that the stress of the cable is uneven or the local aging of the cable material is more serious.
[0037] Furthermore, based on the linear deviation and stress non-uniformity, the first evaluation value is obtained, specifically: The product of the deviation in alignment and the stress non-uniformity is used as the first evaluation value for each cable in each monitoring cycle. It should be noted that the larger the first assessment value, the more significant the change in the mechanical state of the cable, reflecting a more serious abnormality in cable force, and the possibility that the cable may have structural damage, material aging, or local defects.
[0038] Thus, the first evaluation value of each stay cable under each monitoring period is obtained.
[0039] Step 3: Evaluate the distribution dispersion of strain data peaks and their long-term trend in each monitoring period, as well as the deviation of strain data from the linear trend, and calculate the nonlinear intensity of each cable in each monitoring period; calculate the energy mobility of each cable in each monitoring period by using the energy difference of strain data in the frequency domain distribution within the local area where different peaks are located, and determine the second evaluation value of each cable in each monitoring period by combining the nonlinear intensity.
[0040] Furthermore, in the process of monitoring cable force in cable-stayed bridges, relying solely on the first assessment value to evaluate cable force anomalies has certain drawbacks. The lack of assessment of the stability and vibration frequency patterns of strain data at the cable head or anchorage zone may lead to misjudgment and insufficient response capability in monitoring cable force under sudden load conditions, thus affecting the accuracy and reliability of cable force assessment for cable-stayed bridge structures.
[0041] When the cable stays are subjected to sudden loads or damage expansion, the more severe the abnormal changes in cable force caused by these loads, the more unstable the abrupt changes in the strain data of the cable stays, and the more significant the nonlinear long-term trend of the strain data. At the same time, when the abnormal cable force of the cable stays is more likely to be caused by uneven stress or material aging damage, rather than by normal fluctuations in the bridge load, the energy difference of different frequency components in the frequency domain is greater when the strain data is abrupt.
[0042] Based on the above analysis, the nonlinear intensity is calculated by analyzing the nonlinear changes in strain data of each stay cable at different times within each monitoring period, specifically as follows: Obtain the peak values of strain data for each cable at all times within each monitoring cycle; In this embodiment, the AMPD (Automatic Multiscale-based Peak Detection) algorithm is used to obtain the peak. The AMPD algorithm is a well-known technology and will not be described in detail here.
[0043] Calculate the coefficient of variation and Hurst exponent of all peak values for each cable in each monitoring period; It should be noted that the calculation of the coefficient of variation and the Hurst exponent are well-known techniques and will not be elaborated here.
[0044] Linear fitting was performed on the strain data of each cable at all times during each monitoring period, and the fitting error was calculated. In this embodiment, the least squares method is used for linear fitting. The least squares method is a well-known technique and will not be described in detail here. Secondly, the fitting error is measured by calculating the mean absolute error. The calculation of the mean absolute error is a well-known technique and will not be described in detail here.
[0045] The product of the fitting error, the coefficient of variation, and the Hurst exponent is used as the nonlinear strength of each cable in each monitoring period. It should be noted that, in order to avoid the significant mean regression characteristics of the peak-to-peak changes in the strain data, which would cause the Hurst exponent to be 0, the Hurst exponent is positively mapped. The specific positive mapping process is as follows: the sum of the Hurst exponent and a preset value is taken as the result of the positive mapping. The preset value is 0.5. As other implementation methods, implementers can set it according to the actual situation.
[0046] It should be noted that a larger coefficient of variation indicates greater fluctuations in the peak and trough values of the strain data, reflecting poor abrupt change stability of the strain data; a larger Hurst exponent indicates a strong inherent trend in the strain change of the cable, and this trend will continue, reflecting an irreversible deterioration process in the cable force driven by intrinsic factors such as damage propagation, material relaxation, or continuous loss of cable force; a larger fitting error indicates significant nonlinear changes in the cable strain data; and a larger nonlinear intensity indicates that the cable force is in an unstable and continuously deteriorating dangerous state, reflecting a more severe condition of the cable under sudden loads or damage propagation.
[0047] Secondly, the energy difference of the frequency components of the strain data in the frequency domain within the local area corresponding to the wave peak of each cable in each monitoring period is analyzed, and the energy mobility is calculated, specifically: For each cable-stayed cable, under each monitoring cycle, multiple moments within the neighborhood of each peak moment are recorded as local time periods; In this embodiment, the duration of the local time period is 30 seconds. As for other implementation methods, the implementer can set it according to the actual situation.
[0048] Frequency domain analysis was performed on the strain data at all times within a local time period to obtain the spectrum corresponding to each wave peak; In this embodiment, the Fourier transform algorithm is used for frequency domain analysis. The Fourier transform algorithm is a well-known technology and will not be described in detail here.
[0049] The energy of all frequency components within the spectrum is used to form an energy sequence; The sum of the differences in the energy sequences of any two wave peaks for each cable in each monitoring period is calculated as the energy mobility of each cable in each monitoring period. In this embodiment, the sum of the DTW distances of the energy sequences of any two peaks of each stay cable in each monitoring period is calculated as the energy mobility of each stay cable in each monitoring period. The calculation of the DTW distance is a well-known technique and will not be described in detail here.
[0050] It should be noted that the greater the energy mobility, the greater the strain abrupt changes that occur at different times during the entire monitoring period. The underlying vibration frequency components are significantly different, reflecting significant changes in the stress distribution of the cable and the existence of localized stress unevenness or material aging.
[0051] Furthermore, based on the nonlinear intensity and energy mobility, a second evaluation value is determined, specifically as follows: The product of nonlinear intensity and energy mobility is used as the second evaluation value for each cable in each monitoring period; It should be noted that the larger the second evaluation value, the more complex the nonlinear characteristics and long-term trend of the strain data, and the more significant the change in the frequency component energy distribution of the strain data. This reflects that the strain data of the cable-stayed cable exhibits more significant abnormal characteristics, indicating that the cable-stayed cable is more severely affected by sudden loads or damage propagation.
[0052] Thus, the second evaluation value for each stay cable under each monitoring period was obtained.
[0053] Step 4: Based on the first and second evaluation values, determine the anomaly coefficient of each cable in each monitoring period, and conduct anomaly assessment of the cable force state of the cable-stayed bridge.
[0054] During the monitoring of cable tension on cable-stayed bridges, the greater the degree of deviation of the cable's overall alignment in the spatial dimension, the more pronounced the stress unevenness and local aging characteristics in the cable region (i.e., the higher the first evaluation value), and the more severe the sudden load or damage propagation caused by stress unevenness or material aging damage in the time dimension (i.e., the higher the second evaluation value), the more likely the cable tension on the cable-stayed bridge will be abnormal. Therefore, based on the first and second evaluation values, an anomaly coefficient is calculated as follows: A comprehensive evaluation algorithm is used to comprehensively evaluate the first and second evaluation values of each cable in each monitoring period, and obtain the comprehensive score of each cable in each monitoring period as the anomaly coefficient of each cable in each monitoring period. In this embodiment, the TOPSIS (Technique for Order Preference by Similarity to Ideal Solution) algorithm is used for comprehensive evaluation. The TOPSIS algorithm is a well-known technology and will not be described in detail here. Secondly, the weights in the TOPSIS algorithm are calculated using the entropy weight method, which is also a well-known technology and will not be described in detail here.
[0055] It should be noted that the larger the anomaly coefficient, the more severe the cable force anomaly of the stay cable during the entire monitoring period, and the higher the risk that the stay cable is in a highly abnormal state during this monitoring period. The flowchart of the method for obtaining the anomaly coefficient provided in this application embodiment is shown below. Figure 2 As shown.
[0056] Secondly, based on the anomaly coefficient, the cable force state of the stay cables on the cable-stayed bridge is evaluated, specifically as follows: Obtain the segmentation threshold of the anomaly coefficient of all stay cables on the cable-stayed bridge in each monitoring period; In this embodiment, the Otsu's method is used to obtain the segmentation threshold. Otsu's method is a well-known technique and will not be described in detail here. As other implementation methods, implementers may use other methods of the prior art, such as cross-validation. This embodiment does not impose any special restrictions on this.
[0057] Cables with an anomaly coefficient greater than the segmentation threshold are classified as abnormal cables. If each cable is classified as an abnormal cable in multiple consecutive monitoring periods, then the cable force state of the cable is abnormal; otherwise, the cable force state of the cable is not abnormal. In this embodiment, if each cable is an abnormal cable in 5 consecutive monitoring cycles, then the cable tension state of the cable is abnormal.
[0058] For any cable-stayed cable exhibiting abnormal tension, maintenance personnel should be notified promptly for inspection and repair to ensure the safe operation of the cable-stayed bridge.
[0059] Based on the same inventive concept as the above methods, this application also provides a smart monitoring system for cable-stayed bridge cable forces based on multimodal data fusion, including a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of any one of the above-described smart monitoring methods for cable-stayed bridge cable forces based on multimodal data fusion.
[0060] It should be understood that, although Figure 1 The steps in the flowchart are shown sequentially as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order in which these steps are executed, and they can be performed in other orders. Figure 1 At least some of the steps in the process may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least some of the sub-steps or stages of other steps.
[0061] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.
[0062] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application. Therefore, any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of this application, without departing from the content of the technical solution of this application, shall fall within the protection scope of the technical solution of this application.
Claims
1. A smart monitoring method for cable-stayed bridge cable forces based on multimodal data fusion, characterized in that, The method includes the following steps: Acquire grayscale images and strain data of each stay cable on the cable-stayed bridge at each time point within each monitoring cycle; Extract the cable skeleton from the grayscale image; for each cable in each monitoring period, analyze the differences in the bending features of pixels on the cable skeleton in the grayscale image at different times, as well as the differences in the distribution of the overall spatial shape of the cable skeleton, and calculate the linear deviation of each cable in each monitoring period. Based on the fluctuation of gray values of pixels on the cable frame in grayscale images at different times, and the inconsistency of gradient direction distribution of pixels on the cable frame, the stress non-uniformity of each cable in each monitoring period is calculated. Combined with the linear deviation, the first evaluation value of each cable in each monitoring period is obtained. The distribution dispersion of strain data peaks and their long-term trend, as well as the deviation of strain data from linear trends, are evaluated for each cable in each monitoring period. The nonlinear intensity of each cable in each monitoring period is calculated. By analyzing the energy difference in the frequency domain distribution of strain data within the local area where different peaks are located, the energy mobility of each cable in each monitoring period is calculated. Combined with the nonlinear intensity, the second evaluation value of each cable in each monitoring period is determined. Based on the first and second evaluation values, the anomaly coefficient of each cable is determined for each monitoring period, and the anomaly assessment of the cable force state of the cable-stayed bridge is carried out.
2. The intelligent monitoring method for cable-stayed bridge cable force based on multimodal data fusion as described in claim 1, characterized in that, The calculation of the alignment deviation of each stay cable in each monitoring period includes: For each cable in each monitoring period, curve fitting is performed on the position coordinates of all pixels on the cable skeleton in the grayscale image at each time point to obtain the fitted curve, and the mean value of the curvature of all pixels on the fitted curve is calculated as the average curvature; the sum of the differences in the average curvature of the grayscale images at any two times in each monitoring period is calculated. Calculate the sum of the correlation between the grayscale images and the fitted curves for any two times within each monitoring period, and then perform a positive mapping on them; The linear deviation is the ratio of the sum to the result of the positive mapping.
3. The intelligent monitoring method for cable-stayed bridge cable force based on multimodal data fusion as described in claim 1, characterized in that, The calculation of stress non-uniformity for each cable in each monitoring period includes: For each cable in each monitoring period, calculate the dispersion of gray values of all pixels on the cable skeleton in the grayscale image at each time point; calculate the average value of the dispersion of grayscale images at all times in each monitoring period. Obtain the gradient directions of all pixels on the cable-stayed bridge frame and form a gradient sequence; calculate the sum of the JS divergence of the gradient sequences corresponding to any two grayscale images within each monitoring period, and denote it as the gradient difference; The stress nonuniformity is the product of the average value and the gradient difference.
4. The intelligent monitoring method for cable-stayed bridge cable force based on multimodal data fusion as described in claim 1, characterized in that, The first evaluation value is the product of the linear deviation and the stress non-uniformity.
5. The intelligent monitoring method for cable-stayed bridge cable force based on multimodal data fusion as described in claim 1, characterized in that, The calculation of the nonlinear strength of each stay cable in each monitoring period includes: Obtain the peak values of strain data for each cable at all times during each monitoring period; calculate the coefficient of variation and Hurst exponent of the peak values of all peak values for each cable during each monitoring period. Linear fitting was performed on the strain data of each cable at all times during each monitoring period, and the fitting error was calculated. The nonlinear intensity is the product of the fitting error, the coefficient of variation, and the Hurst exponent, which are positively mapped together.
6. The intelligent monitoring method for cable-stayed bridge cable force based on multimodal data fusion as described in claim 5, characterized in that, The calculation of the energy mobility of each cable in each monitoring cycle includes: For each cable, under each monitoring cycle, multiple moments in the neighborhood of each wave peak are recorded as local time periods; frequency domain analysis is performed on the strain data of all moments within the local time period to obtain the spectrum corresponding to each wave peak, and the energy of all frequency components in the spectrum is used to form an energy sequence. The energy mobility is the sum of the differences in the energy sequences of any two peaks for each cable in each monitoring period.
7. The intelligent monitoring method for cable-stayed bridge cable force based on multimodal data fusion as described in claim 1, characterized in that, The second evaluation value is the product of nonlinear intensity and energy mobility.
8. The intelligent monitoring method for cable-stayed bridge cable force based on multimodal data fusion as described in claim 1, characterized in that, The determination of the anomaly coefficient of each cable in each monitoring period includes: using a comprehensive evaluation algorithm to comprehensively evaluate the first evaluation value and the second evaluation value of each cable in each monitoring period, and obtaining the comprehensive score of each cable in each monitoring period as the anomaly coefficient of each cable in each monitoring period.
9. The intelligent monitoring method for cable-stayed bridge cable force based on multimodal data fusion as described in claim 1, characterized in that, The method for anomaly assessment of the cable force state of the cable-stayed bridge includes: obtaining the segmentation threshold of the anomaly coefficient of all the cable-stayed bridge in each monitoring period; recording the cable-stayed bridge with anomaly coefficient greater than the segmentation threshold as an abnormal cable-stayed bridge; if each cable-stayed bridge is an abnormal cable-stayed bridge in multiple consecutive monitoring periods, then the cable force state of the cable-stayed bridge is abnormal, otherwise, the cable force state of the cable-stayed bridge is not abnormal.
10. A smart monitoring system for cable-stayed bridge cable forces based on multimodal data fusion, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent monitoring method for cable-stayed bridge cable force based on multimodal data fusion as described in any one of claims 1-9.