Bridge modal parameter identification method based on visual partition measurement

Abstract

The application discloses a bridge modal parameter identification method based on visual partition measurement, comprising the following steps: 1. collecting a checkerboard image under a region of interest, identifying feature points of the checkerboard image, calculating a camera intrinsic parameter matrix and performing camera correction; 2. performing vibration testing on the region of interest, collecting a structure vibration video containing a physical target under the region of interest, and extracting a pixel displacement signal of the structure region of interest through an L-K optical flow method; 3. calculating the camera intrinsic parameter matrix and a rotation and translation matrix, performing three-dimensional reconstruction on the pixel displacement signal, and obtaining a real displacement signal; 4. obtaining an acceleration signal through the displacement signal, and obtaining a vibration mode of the region of interest through a random subspace method; and 5. obtaining an overall vibration mode of the structure after vibration mode splicing is performed on the vibration modes of all the regions of interest. The application can perform non-contact displacement measurement on a bridge structure and obtain modal parameters of the bridge structure, and has the advantages of cost saving, convenience and rapidness.

Description

Technical Field

[0001] This invention relates to the field of bridge structural health monitoring, specifically a method for identifying bridge modal parameters based on visual zoning measurements. Background Technology

[0002] During the operation of bridge structures, the safety, serviceability, and durability inevitably decline under the influence of various loads. If these structures are not monitored in a timely manner, it may lead to major safety accidents. Therefore, health monitoring of bridge structures during their operational phase is essential. Traditional methods for measuring bridge structural deformation mainly include contact and non-contact methods. Contact methods require sensors to be pre-positioned on the bridge structure. In recent years, with the development of measurement technology, high-precision, long-distance, non-contact bridge deformation measurement methods have been proposed. However, these traditional sensors have significant drawbacks. For example, traditional contact sensors are usually placed at critical cross-sections of the structure, only acquiring monitoring data from a limited number of discrete measurement points; to obtain high-resolution modal vibrations, a large number of sensors need to be installed, which is costly and wasteful of manpower, but the additional sensor weight introduces a certain degree of uncertainty. Non-contact sensors, on the other hand, cannot track fixed points on the surface of an object, and may lose tracking if the amplitude is too large.

[0003] Besides using sensors to acquire measured data of engineering structures, computer vision-based measurement techniques using images or video frames can be applied to structural displacement and vibration measurements. However, when measuring the global field of view of a bridge structure, the high-frequency vibration amplitude at the measurement point is very small. Computer vision is not adept at detecting minute structural vibrations, requiring more field-of-view information to be obtained within limited image resolution. Therefore, there is an urgent need for a camera with a wider field of view and higher resolution, but this would significantly increase monitoring costs. Summary of the Invention

[0004] The present invention aims to address the shortcomings of the existing technology by proposing a method for identifying bridge modal parameters based on visual partitioning measurement, in order to improve measurement accuracy, control monitoring costs, and thus achieve accurate identification of bridge modal vibration modes.

[0005] To achieve the above-mentioned objectives, the present invention adopts the following technical solution:

[0006] The present invention provides a method for identifying bridge modal parameters based on visual partitioning measurements, characterized by the following steps:

[0007] Step 1: Install and calibrate the data acquisition devices at each measuring point on the bridge;

[0008] Step 2: Use the calibrated acquisition devices to capture vibration videos of the bridge in all field-of-view zones and perform video distortion correction on the videos to obtain the distortion-corrected videos;

[0009] Step 3: Select the region of interest (ROI) in each frame of the distortion-corrected video under each partition field of view, and detect the feature points in the ROI of each frame of the video using the Shi-Tomasi corner detection algorithm. Select key feature points from them, and then use the LK optical flow method to track the motion process of the key feature points in each frame of the video to obtain the relative displacement of key feature points in adjacent frames of the video under each partition field of view. Obtain the pixel displacement time history curve of each key feature point in the ROI of the video under each partition field of view, and use it as the pixel displacement signal of the ROI.

[0010] Step 4: Reconstruct the pixel displacement signal to obtain the actual displacement time history curve in three-dimensional space, and use it as the real displacement signal. Process the real displacement signal by numerical differentiation method to obtain the actual acceleration time history curve of each feature point in the region of interest in the video under each partition field of view, and use it as the real acceleration signal.

[0011] Step 5: Based on the actual acceleration signal, the mode shape of the region of interest in the video under each partition field of view is calculated using the random subspace method and then stitched together to obtain the complete mode shape of the bridge structure.

[0012] The bridge modal parameter identification method based on visual partitioning measurement described in this invention is characterized in that the mode shapes of the region of interest in each partitioned field of view in step 5 are stitched together according to the following process:

[0013] Step a: Take the coincidence point of adjacent regions of interest in the video under each partition field of view as the reference point, calculate the ratio between the reference point and its corresponding mode shape under each partition field of view, and use it as the scaling factor under each partition field of view;

[0014] Step b: Calculate the product of the mode shape of the region of interest in the video under each partition field of view and its scaling factor, thereby scaling the mode shape under each partition field of view to obtain the scaled mode shape under each partition field of view;

[0015] Step c: Keep the mode shape of the region of interest in the video under the first partition field of view unchanged, and then stitch together the remaining scaled mode shapes to obtain the overall mode shape of the bridge.

[0016] The present invention provides an electronic device, including a memory and a processor, wherein the memory is used to store a program that supports the processor in executing the bridge modal parameter identification method, and the processor is configured to execute the program stored in the memory.

[0017] The present invention discloses a computer-readable storage medium on which a computer program is stored, wherein the computer program, when executed by a processor, performs the steps of the bridge modal parameter identification method.

[0018] Compared with the prior art, the beneficial effects of the present invention are reflected in:

[0019] 1. Compared with the traditional measurement method of arranging a large number of sensors on the bridge, the present invention realizes non-contact measurement, which is convenient to operate and saves manpower and resources.

[0020] 2. This invention obtains the vibration response of each zone through visual zone measurement. Compared with measuring the entire bridge under a global field of view, it addresses the problem of limited field of view and controls monitoring costs.

[0021] 3. This invention uses a fixed-axis automatic rotation device to acquire images in each zone. Compared with multi-point moving measurement methods, it can solve the problem of difficult data acquisition of actual bridges, and is easy to operate and highly convenient.

[0022] 4. This invention, through theoretical derivation, restores the pixel displacement time history curve to the actual displacement time history curve, which, compared with the traditional scaling factor method, addresses the accuracy problem of response restoration at different positions. Attached Figure Description

[0023] Figure 1 This is a flowchart of the bridge modal parameter identification method based on visual partitioning measurement according to the present invention;

[0024] Figure 2 This is a schematic diagram of the bridge modal parameter identification and device based on visual partition measurement according to the present invention.

[0025] Figure 3 This is a diagram of the camera imaging model of the present invention;

[0026] Figure 4 This is a schematic diagram of the mode splicing of the present invention;

[0027] Figure 5 This is a laboratory test diagram of the simply supported beam bridge model of the present invention;

[0028] Figure 6 This is a comparison diagram of the displacement response identified at point S1 and the displacement response acquired by the sensor, as presented in this invention.

[0029] Figure 7 This is a comparison diagram of the displacement response identified at point S2 and the displacement response acquired by the sensor, as presented in this invention.

[0030] Figure 8 This is a comparison diagram of the displacement response at point S3 identified by the present invention and the displacement response acquired by the sensor;

[0031] Figure 9 This is a comparison diagram of the first-order vibration mode of the bridge splicing according to the present invention and the actual vibration mode of the bridge;

[0032] Figure 10 This is a comparison diagram of the second-order vibration mode of the bridge splicing according to the present invention and the actual vibration mode of the bridge. Detailed Implementation

[0033] In this embodiment, a bridge modal parameter identification method based on visual partitioning measurement is described, referencing... Figure 1 , Figure 2 This includes the following steps:

[0034] Step 1: Install and calibrate the data acquisition devices at each measuring point on the bridge;

[0035] In this embodiment, 10-20 images of the checkerboard pattern need to be captured from multiple angles, using highly recognizable black and white checkerboard images. During image acquisition, autofocus needs to be disabled, and manual focus performed to ensure the camera's focal length remains constant throughout the shooting process. All images are then input into MATLAB and calibrated using Camera Calibrator to obtain the camera's intrinsic parameter matrix. And the lens distortion coefficients (k1,k2,k3,p1,p2), where f is the physical focal length of the camera, x0 and y0 are the coordinates of the intersection of the camera's optical axis and the image plane in the camera coordinate system, dX and dY are the physical dimensions of a single pixel on the X and Y axes, respectively, k1 is the quadratic radial distortion coefficient, k2 is the quartic radial distortion coefficient, k3 is the sixth radial distortion coefficient, p1 is the quadratic tangential distortion coefficient on the horizontal axis, and p2 is the quadratic tangential distortion coefficient on the vertical axis.

[0036] Step 2: Use the calibrated acquisition devices to capture vibration videos of the bridge in all field-of-view zones and perform video distortion correction on the videos to obtain the distortion-corrected videos;

[0037] In this embodiment, a single-point excitation method is used to excite the experimental beam, and vibration videos of this bridge section are captured. A computer equipped with an OpenCV-Python environment platform is used to correct the distortion of the original video using distortion coefficients (k1,k2,k3,p1,p2).

[0038] Assume (x, y) is the original position of the distorted point on the image, and r is the distance of that point from the center of the imager, i.e., r 2 =x 2 +y 2 , These are the pixel coordinates of the distorted image. Radial distortion can be corrected using equations (1) and (2):

[0039]

[0040]

[0041] Tangential distortion occurs when the image plane and the lens are not perfectly parallel. Tangential distortion can be corrected using equations (3) and (4):

[0042]

[0043]

[0044] Step 3: Select the region of interest (ROI) in each frame of the distortion-corrected video under each partition field of view, and detect the feature points in the ROI of each frame of the video using the Shi-Tomasi corner detection algorithm. Select key feature points from them, and then use the LK optical flow method to track the motion process of the key feature points in each frame of the video to obtain the relative displacement of key feature points in adjacent frames of the video under each partition field of view. Obtain the pixel displacement time history curve of each key feature point in the ROI of the video under each partition field of view, and use it as the pixel displacement signal of the ROI.

[0045] In this embodiment, a computer equipped with an OpenCV-Python environment platform is used to detect corners within the region of interest using the Shi-Tomasi corner detection algorithm. Points near the sensor are selected as tracking points for the LK optical flow method. Since the optical flow method obtains the relative displacement of feature points in adjacent frames, the data at each time step needs to be accumulated to obtain the pixel displacement signal disp of the key feature point at any time step.

[0046] Step 4: Reconstruct the pixel displacement signal to obtain the actual displacement time history curve in three-dimensional space, and use it as the real displacement signal. Process the real displacement signal by numerical differentiation method to obtain the actual acceleration time history curve of each feature point in the region of interest in the video under each partition field of view, and use it as the real acceleration signal.

[0047] In this embodiment, the establishment of the camera imaging model mainly involves three coordinate systems: image coordinate system, camera coordinate system, and world coordinate system. Computer vision obtains the actual position of the object through transformations between different coordinate systems. Figure 3 This is a camera imaging model.

[0048] The coordinate mapping relationship between the image physical coordinate system and the image pixel coordinate system:

[0049]

[0050] In equation (5), coordinates x and y represent the column number and row number of a certain pixel in the image matrix, respectively. Among them, x0 and y0 are the coordinates of the intersection of the camera optical axis and the image plane in the camera coordinate system, coordinate system oxy is the image pixel coordinate system, and coordinate system OXY is the image physical coordinate system.

[0051] Coordinate mapping relationship between the image physical coordinate system and the camera coordinate system:

[0052]

[0053] In equation (6), f is the physical focal length of the camera, and the coordinate system is O. c X c Y c Z c Let be the camera coordinate system.

[0054] Coordinate mapping relationship between camera coordinate system and world coordinate system:

[0055]

[0056] In equation (7), R is the orthogonal unit rotation matrix, T is the three-dimensional translation vector, and the coordinate system O is... w X w Y w Z w Use the world coordinate system.

[0057] Using the relationships between the above coordinates, the coordinates of a point in the world coordinate system can be transformed into image pixel coordinates, as shown in the following formulas:

[0058]

[0059] Let the plane containing the chessboard grid be Z in the world coordinate system. w = 0 in the plane, so the world coordinates of any corner point P of the chessboard are (X = 0). w ,Y w According to equation (8), the world coordinates (X, 0) are obtained. w ,Y w The relationship between (x, y) and pixel coordinates (x, y) is as follows:

[0060]

[0061] The selected feature point is used as the compensation point, with its pixel coordinates set to (u,v) and its mapped world coordinates set to (U,V). The feature point pixel displacement disp is then superimposed with (u,v) to obtain the result. Its world coordinates are The actual displacement of the feature point can be obtained by subtracting it from the world coordinates (U,V) of the compensation point. The true acceleration signal of the feature point is obtained by numerical differentiation.

[0062] Step 5: Based on the actual acceleration signal, the mode shape of the region of interest in the video under each partition field of view is calculated using the random subspace method and then stitched together to obtain the complete mode shape of the bridge structure.

[0063] In this embodiment, a vertical acceleration signal is selected. The frequency and mode shape of the region of interest are obtained by using the random subspace method for time-domain processing and analysis.

[0064] In practice, the mode shapes of the regions of interest in each partition's field of view in step 5 are stitched together according to the following process:

[0065] Step a: Take the coincidence point of adjacent regions of interest in the video under each partition field of view as the reference point, calculate the ratio between the reference point and its corresponding mode shape under each partition field of view, and use it as the scaling factor under each partition field of view;

[0066] Step b: Calculate the product of the mode shape of the region of interest in the video under each partition field of view and its scaling factor, thereby scaling the mode shape under each partition field of view to obtain the scaled mode shape under each partition field of view;

[0067] Step c: Keep the mode shape of the region of interest in the video under the first partition field of view unchanged, and then stitch together the remaining scaled mode shapes to obtain the overall mode shape of the bridge.

[0068] In this embodiment, let the mode shapes of all regions of interest be Md1(a1,a2,a3), Md2(a4,a5,a6), ..., where a3 and a4 are the modal values ​​of the coincident points of adjacent segments, and the scaling factor is denoted as ... Then Md2'=γ 1,2 ×Md2, further, adjacent vibration modes can be spliced ​​together Md 1,2 =(a1,a2,a3,γ) 1,2 a5,γ 1,2 a6), similarly Each mode shape can be spliced ​​together to obtain the overall mode shape Md of the bridge, such as... Figure 4 As shown.

[0069] In this embodiment, an electronic device includes a memory and a processor. The memory stores a program that supports the processor in executing the above-described method, and the processor is configured to execute the program stored in the memory.

[0070] In this embodiment, a computer-readable storage medium stores a computer program, which is executed by a processor to perform the steps of the above method.

[0071] The following describes the bridge modal parameter identification method and device based on visual partitioning measurement according to the present invention with specific examples. In the experimental verification, the method employed... Figure 5 The diagram shows the sensor arrangement of the bridge structure. Figure 5 The parameters of the bridge under test are set as follows: bridge span length is 300cm, width is 12.5cm, height is 2cm, and bridge density is 2700kg / m³. 3 The camera sampling frequency is 60Hz. Accelerometers were deployed at spans 1 / 8, 1 / 4, 3 / 8, 1 / 2, 5 / 8, 3 / 4, and 7 / 8 of the bridge to acquire acceleration responses for analysis of the bridge's natural frequencies and mode shapes. Displacement sensors were deployed at spans 3 / 8, 1 / 2, and 5 / 8 of the bridge to acquire displacement responses for verification of response identification accuracy. The identified displacement responses were compared with the sensor-acquired displacement responses. Figure 6 , Figure 7 and Figure 8 As shown in the figure, the displacement identification results at different points obtained by this method are in good agreement. Modal parameter identification was performed on the acceleration response reconstructed from the displacement response extracted from the vibration videos collected from each zone. The identified natural frequencies of the structure were compared with the bridge vibration modes identified from the data collected by the acceleration sensors, as shown in Table 1.

[0072] Table 1

[0073]

[0074] As shown in Table 1, this method accurately identifies the natural frequencies. The overall modal vibration mode of the bridge, obtained by stitching together the identified mode shapes of each zone, is compared with the bridge mode shapes identified from the data collected by the acceleration sensor. Figure 9 and Figure 10 As shown, the vibration modes identified by this method are highly similar to those identified by the sensor.

Claims

1. A method for identifying bridge modal parameters based on visual partitioning measurement, characterized in that, Includes the following steps: Step 1: Install and calibrate the data acquisition devices at each measuring point on the bridge; Step 2: Use the calibrated acquisition devices to capture vibration videos of the bridge in all field-of-view zones and perform video distortion correction on the videos to obtain the distortion-corrected videos; Step 3: Select the region of interest (ROI) in each frame of the distortion-corrected video under each partition field of view, and detect the feature points in the ROI of each frame of the video using the Shi-Tomasi corner detection algorithm. Select key feature points from them, and then use the LK optical flow method to track the motion process of the key feature points in each frame of the video to obtain the relative displacement of key feature points in adjacent frames of the video under each partition field of view. Obtain the pixel displacement time history curve of each key feature point in the ROI of the video under each partition field of view, and use it as the pixel displacement signal of the ROI. Step 4: Reconstruct the pixel displacement signal to obtain the actual displacement time history curve in three-dimensional space, and use it as the real displacement signal. Process the real displacement signal by numerical differentiation method to obtain the actual acceleration time history curve of each feature point in the region of interest in the video under each partition field of view, and use it as the real acceleration signal. Step 5: Based on the actual acceleration signal, the mode shape of the region of interest in the video under each partition field of view is calculated using the random subspace method and then stitched together to obtain the complete mode shape of the bridge structure.

2. The method for identifying bridge modal parameters based on visual partitioning measurement according to claim 1, characterized in that, The mode shapes of the regions of interest in each partitioned field of view in step 5 are stitched together according to the following process: Step a: Take the coincidence point of adjacent regions of interest in the video under each partition field of view as the reference point, calculate the ratio between the reference point and its corresponding mode shape under each partition field of view, and use it as the scaling factor under each partition field of view; Step b: Calculate the product of the mode shape of the region of interest in the video under each partition field of view and its scaling factor, thereby scaling the mode shape under each partition field of view to obtain the scaled mode shape under each partition field of view; Step c: Keep the mode shape of the region of interest in the video under the first partition field of view unchanged, and then stitch together the remaining scaled mode shapes to obtain the overall mode shape of the bridge.

3. An electronic device, comprising a memory and a processor, characterized in that, The memory is used to store a program that supports the processor in executing the bridge modal parameter identification method of claim 1 or 2, and the processor is configured to execute the program stored in the memory.

4. A computer-readable storage medium storing a computer program thereon, characterized in that, The computer program, when run by a processor, executes the steps of the bridge modal parameter identification method according to claim 1 or 2.

Images