Reconstruction method for simulating local scouring terrain of pier in sandpit environment
By moving the camera on a circular track and combining it with the VisualSFM and PMVS algorithms, the difficult problem of riverbed scour topography reconstruction was solved, efficient and accurate simulation of local scour topography of bridge piers was achieved, and high-precision riverbed evolution analysis data was provided.
Patent Information
- Application Number
- CN202510777551.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-11
- Publication Date
- 2025-10-17
AI Technical Summary
Existing technologies make it difficult to efficiently and accurately reconstruct riverbed scour topography, especially the simulation and testing of local scour topography of bridge piers in sand mining environments. There is a lack of full-coverage, non-contact, fast and efficient measurement methods.
A mobile camera was used on a circular track to capture the riverbed topography. The VisualSFM platform and PMVS algorithm were combined to generate a dense 3D point cloud. The terrain was reconstructed through gridding and interpolation methods, and the area and depth of local scour pits on the bridge piers were quantitatively calculated.
It achieves fast, efficient and accurate three-dimensional reconstruction of riverbed terrain, provides high-precision riverbed evolution analysis, and provides reliable data support for bridge pier scour research.
Smart Images
Figure CN120797586A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to a reconstruction method for simulating the local scour topography of a bridge pier in a sand pit environment. BACKGROUND
[0002] The construction of bridges enhances the local turbulence of water flow, destroys the balance of sediment movement and the original riverbed morphology. With the increasing demand for modern building materials, river sand mining has become an important source of engineering sand. Riverbed sand mining can exacerbate changes in riverbed structure, threaten bridge safety, and even cause bridge damage. Once the sand pit is formed, it will exist, develop and evolve for a long time, intensifying the scouring intensity of water flow around the riverbed and bridge piers. Therefore, it is of great significance to study accurate and efficient three-dimensional terrain measurement methods for river model tests and practical engineering applications.
[0003] Riverbed evolution is the result of sediment movement. Many scholars have studied how to efficiently and accurately obtain river model test terrain data. In the past, most terrain data was obtained through traditional single-point measurement (leveling instrument, theodolite, total station, measuring needle, etc.). New measurement methods such as laser scanners and ultrasonic terrain meters have improved in practicality. Overall, there are fewer full-coverage non-contact, fast, efficient, and low-cost riverbed terrain measurement methods, and there is a lack of analysis of scour topography changes.
[0004] With the rapid development of photogrammetry and computer vision technology, it is possible to reconstruct three-dimensional terrain through camera technology, which provides a possibility for river model test terrain research. However, there are still unresolved scientific problems in combining image three-dimensional reconstruction technology with bridge pier scour topography in river model tests with sand pits. SUMMARY
[0005] In order to more accurately simulate and test the environment terrain of the bridge pier in the river with a sand pit environment, the technical scheme of the present application is as follows:
[0006] A reconstruction method for simulating the local scour topography of a bridge pier in a sand pit environment, comprising a bridge pier local scour experimental model, the bridge pier local scour experimental model comprising a water tank, the bottom of the water tank being paved with sand and gravel, a vertical support rod simulating a bridge pier being fixed in the middle of the water tank, the vertical support rod having a recessed sand pit area on the side thereof; an annular track being arranged above the water tank, a moving camera being arranged on the annular track and being capable of moving along the annular track; the water tank having a water inlet at one end and a drain at the other end;
[0007] Step (1): In the shooting area of the annular area to be tested, black and white checkerboards are first prepared as terrain reconstruction control points, and the terrain reconstruction control points are evenly arranged;
[0008] Step (2): Conduct a flume riverbed scour model test, release water from one side of the flume, and control the water flow rate and time according to the design requirements; then turn off the water flow, slowly reduce the flume flow to a stop, and use a mobile camera to take circular photos of the study area when there is no water accumulation in the study area;
[0009] Step (3): The photos are input into the VisualSFM platform for calculation, and the coordinates of the set terrain reconstruction control points are input into the model. Feature points of multiple adjacent photos taken continuously are matched to identify the same feature points. On this basis, the PMVS algorithm is used to optimize the results of the sparse 3D point cloud reconstruction, that is, to generate a dense 3D point cloud of the scour bed surface.
[0010] Step (4): Divide the sand-topped area into a grid of square areas, and use the interpolation method to process the point cloud data in the grid to obtain the interpolation results on the grid points;
[0011] Step (5): Analyze the topographic characteristics of the local scour bed of the bridge pier when there is a sand pit and quantitatively calculate the pit area:
[0012] Step (6): After the area to be measured is gridded, the maximum pit depth and pit area are calculated.
[0013] Preferably, the quantitative calculation of the pit area in step (5) is performed as follows: first, the slope elevation is subtracted to level the calculation domain. The specific steps are as follows:
[0014] Assume that the slope of the water tank is J = tanθ, there are n grids in the X direction, and the height of the slope at the i-th grid is h i =J(net n -net i );
[0015] Subtract the height of the tilt h from the height of the i-th grid i That is the terrain elevation when it is a horizontal plane, that is, the calculation domain is horizontal.
[0016] Preferably, in step (6), the area of the pier scour pit at different depths is quantitatively calculated, and the maximum pit depth is calculated by selecting the plane near the central axis of the sand mining pit and the plane near the symmetry axis of the pier model through meshing. 坑 The N lowest elevation points Z processed i The pit area is calculated based on the corresponding area of the maximum pit depth, the pit depth d s , pit area V 坑 The calculation formula is as follows:
[0017]
[0018] Preferably, the adjacent photo overlap rate reaches 80% or above.
[0019] Preferably, the annular track is fixed on a support, the support is fixed above a water tank, the vertical support rod is located in the middle of the water tank, the lower part of the vertical support rod is embedded in the gravel layer, and the annular track above the water tank is axially coincident with the axis of the vertical support rod.
[0020] Preferably, the annular track is provided with a moving vehicle capable of moving along the track, the moving vehicle has a lower part with a ring-shaped track matched with the annular track and a ring-shaped tooth surface outside the annular track, a driving motor is arranged on the moving vehicle, a gear engaged with the tooth surface is fixed to the output end of the driving motor, the driving motor is driven to move along the annular track through the gear, and the moving camera is fixed to the moving vehicle.
[0021] Preferably, in order to simulate the river flow state or simulate different water flow velocities, one end of the water tank is provided with a water inlet pipe, the other end is provided with a water outlet pipe, a flow valve is arranged at the water inlet pipe, and a filter is arranged at the water outlet pipe.
[0022] Compared with the prior art, the present application has the following beneficial effects:
[0023] The present application is applicable to both water tank tests and riverbed physical model tests, can quickly and efficiently reconstruct the three-dimensional topography of the riverbed before and after scouring, has high precision, provides high-precision simulation and calculation for studying riverbed evolution and carrying out water conservancy model tests, and is beneficial to the research of bridge pier scouring topography. BRIEF DESCRIPTION OF DRAWINGS
[0024] Figure 1 The present application is a chessboard grid control point schematic diagram.
[0025] Figure 2 The present application is a bridge pier local scouring experimental model structure schematic diagram.
[0026] Figure 3 The present application is a moving camera shooting process diagram and a feature point matching schematic diagram of two consecutive pictures.
[0027] Figure 4 The present application is a topography reconstruction process diagram.
[0028] Figure 5 The present application is a calculation domain horizontalization schematic diagram.
[0029] Figure 6 The present application is a two-dimensional cross-sectional view of the sand pit and the bridge pier model at different distances.
[0030] Figure 7 The present application is a sand pit area cross-sectional view of the sand pit and the bridge pier model at different distances.
[0031] Figure 8The bridge pier scour pit area section view at different distances between the sand pit and the bridge pier model.
[0032] Figure 9 For Figure 2 The local structure enlarged view.
[0033] In the figure: 10-water tank, 110-support, 120-sand layer, 130-water inlet pipe, 140-drainage pipe, 20-sand pit area, 30-vertical support rod, 40-ring track, 410-tooth surface, 50-moving vehicle, 510-driving motor, 520-gear, 60-moving camera, 101-bridge pier model section, 102-scour topography section, 103-sand pit section, 104-bridge pier scour pit section. DETAILED DESCRIPTION
[0034] The application will be described in detail below in conjunction with the drawings and specific embodiments.
[0035] Referring to Figures 1-8 , the method for fine reconstruction analysis of the local scour three-dimensional topography of the bridge pier in the presence of the sand pit combined with the mobile measurement technology will be described in detail below in conjunction with the drawings and embodiments. The method includes the following steps using a bridge pier local scour experimental model:
[0036] The bridge pier local scour experimental model includes a water tank 10, the bottom of which is paved with sand to form a sand layer 120, and a vertical support rod 30 simulating a bridge pier is fixed in the middle of the water tank, and the vertical support rod has a recessed sand pit area 30 beside it; a ring track is arranged above the water tank, and a moving camera 60 capable of moving along the ring track 40 is arranged on the ring track; the water tank has a water inlet pipe at one end and a drainage port at the other end.
[0037] In an embodiment of the application, the ring track 40 is fixed on the support 110, the support 110 is fixed above the water tank, the sand is paved to form a sand layer for simulating the riverbed sand surface, the vertical support rod 30 is located in the middle of the water tank for simulating the bridge pier model, the lower part of the vertical support rod 30 is buried in the sand layer, the sand pit area 30 is used to simulate the sand pit, and the distance between the sand pit area 20 and the vertical support rod 30 can be adjusted to simulate the influence of different sand pits on the local scour of the bridge pier.
[0038] In this embodiment, the axis of the annular track above the water tank coincides with the vertical support rod 30, the annular track 40 is fixed on the upper part of the water tank through the support frame, a movable trolley 50 capable of moving along the track is arranged on the annular track 40, the movable trolley has a lower part with a mating annular track and a tooth surface 410 outside the annular track, a driving motor 510 is arranged on the movable trolley, a gear 520 engaged with the tooth surface 410 is fixed to the output end of the driving motor, the driving motor 510 is driven to move along the annular track through the gear, and the moving camera 60 is fixed on the movable trolley, so that the moving camera can be used to take pictures around the vertical support rod 30 in the annular direction.
[0039] In order to simulate the state of river flow or simulate different water flow velocities, one end of the water tank is provided with a water inlet pipe 130 and the other end is provided with a water outlet pipe 140; the water inlet pipe is provided with a flow valve to control the water outlet amount, or a drainage device is used to control the uniformity of flow, and a filter is arranged at the water outlet pipe to block the sand, and the water outlet pipe 140 is blocked.
[0040] The vertical support rod 30 of the experimental device is located in the middle section of the entire water tank, and the lower end of the vertical support rod is embedded in the sand layer.
[0041] The specific operation steps are as follows:
[0042] Step (1): In the shooting area of the annular area to be tested, a black and white checkerboard is first set as a terrain reconstruction control point, and the terrain reconstruction control points are uniformly arranged;
[0043] A black and white checkerboard is first set as a terrain reconstruction control point (GCPs, Ground Control Point), and the number and arrangement of the checkerboards can be flexibly and uniformly arranged according to the test site. As shown in the figure, uniformly arranging multiple checkerboard control points can enhance the recognition of the moving camera and improve the accuracy of the reconstructed point cloud. Figure 1
[0044] The coordinates are set as A(-0.03, 0, 0.255), B(-0.03, 0, 0.255), C(0.03, 0, 0.255), D(-0.03, 0.03, 0.255), E(0, 0, 0.255), F(0.03, 0.03, 0.255), G(-0.03, 0.06, 0.255), H(0, 0.06, 0.255), I(0, 0.06, 0.255), and the unit is meter.
[0045] Step (2): The water tank bed scouring model test is carried out, water is discharged from one side of the water tank, and the water flow velocity and time are controlled according to the design requirements; then the water flow is closed, the water flow of the water tank is slowly reduced to stop, and when there is no water in the research area, the moving camera is used to take pictures in the annular direction in the research area;
[0046] During the shooting process, the mobile camera 60 is driven by the mobile vehicle 50 to move in a circular direction.
[0047] In the embodiment of the present invention, the overlapping rate of adjacent photos reaches more than 80%, and the more photos containing control points are taken, the higher the accuracy is.
[0048] The project uses SIFT (Scale Invariant Feature Transform) feature vector image technology, which is generated through the following four steps: ① Detect extreme points in scale space and preliminarily determine the location and scale of key points. ② Accurately determine the location and scale of key points, and eliminate uncertain edge points and low-contrast key points. ③ Specify parameters for the key point direction. ④ Generate SIFT feature vectors. Using the kd-trees method, known points can quickly and efficiently find the nearest d-dimensional space point for feature point matching.
[0049] Step (3): The photos are input into the VisualSFM platform for calculation, and the set coordinates are input into the model. Feature points of multiple adjacent photos taken continuously are matched to identify the same feature points. On this basis, the PMVS algorithm is used to optimize the results of the sparse 3D point cloud reconstruction, that is, to generate a dense 3D point cloud of the scour bed surface.
[0050] The VisualSFM platform is a free application for 3D reconstruction of moving structures using a graphical user interface (GUI). It utilizes multi-core parallel processing for feature point detection, feature point matching (with a sophisticated matching algorithm), and bundle adjustment, resulting in rapid operation. Because the 3D point cloud generated by the VisualSFM platform only uses the image coordinate system and lacks camera position, scale, and orientation information, ground control points (GCPs) must be input into the model to convert between the image coordinate system and the world coordinate system, ultimately resulting in a 3D reconstruction in a manually defined coordinate system.
[0051] like Figure 3 a, and 3b show two pictures taken consecutively. Figure 3 aThere are more than 6,000 feature points identified. Figure 3 There are more than 7,000 feature points identified by b, and there are nearly 4,000 similar points after matching. Figure 3 As shown in c.
[0052] The point cloud outside the target area is eliminated, and the XY plane (horizontal direction) boundary is intercepted using the getdata software. This is especially applicable when facing irregular target areas to obtain a dense point cloud of the target area.
[0053] On this basis, the PMVS (Patch based multi-view stereopsis) algorithm is used to optimize the results of sparse three-dimensional point cloud reconstruction, that is, to generate a dense three-dimensional point cloud of the scouring bed surface. The PMVS algorithm uses the Harris and DoG (Difference of Gaussian) algorithm to match and extract feature points of the sparse point cloud, and gradually diffuses the surface after reconstruction of the sparse point cloud as a seed patch to obtain dense patches of the target surface, and finally completes the reconstruction of the scene surface.
[0054] As shown in Figure 4 b, the obtained terrain point cloud is dense and uniformly distributed, showing good reconstruction effect.
[0055] During the experiment, the smaller the research area, the higher the camera pixel, the larger the number of photos, and the higher the accuracy.
[0056] Step (4): The sanding range is divided into a grid of square areas, and an interpolation method is used to process the point cloud data in the grid to obtain the interpolation results on the grid points;
[0057] In order to quantitatively analyze the scouring morphological feature parameters, the research area needs to be gridded, and the point cloud elevation in the grid needs to be interpolated.
[0058] The sanding range of the flume is divided into a grid of 1mm×1mm, and an interpolation method is used to process the point cloud data in the grid to obtain the interpolation results on the grid points, as shown in Figure 4 c and Figure 4 d respectively show the three-dimensional terrain contour maps before and after scouring. After three-dimensional reconstruction of the scouring bed surface, part of the three-dimensional point cloud near the pier column remains. In order to avoid the overlap of the elevation of the pier column model structure and the elevation of the riverbed being included in the interpolation calculation, the irregular pier column model point cloud needs to be removed.
[0059] In this embodiment, the use of a 1mm×1mm grid can control the error of the reconstructed three-dimensional terrain structure within ±0.1mm, with high accuracy. From Figure 4 a, Figure 4 b and Figure 4 c can be seen that the three-dimensional point cloud reconstructed after gridding and interpolation processing is consistent with the actual arrangement of the structure.
[0060] Step (5) is the quantitative calculation of the local scouring bed surface terrain feature of the pier and the area of the sand pit:
[0061] The specific way of the quantitative calculation is to subtract the slope elevation first, and to level the calculation domain, as shown in Figure 5As shown, the specific steps are as follows: Assume that the slope of the water tank is J = tanθ, set the positive direction of the X axis as the direction of water flow, there are n grids in the x direction, and at the i-th grid, the slope height h i =J(net n -net i ), subtract the height of the tilt h from the height of the i-th grid i That is the terrain elevation when it is horizontal, that is, the horizontal revision of the calculation domain.
[0062] Step (5): After the area to be measured is gridded, the maximum pit depth and pit area are calculated.
[0063] In order to study the structural characteristics of the scour bed, based on the Cartesian coordinate system, dimensionless elevation profiles were drawn along the longitudinal XZ section of the centerline plane of the flume length and the transverse YZ section of the symmetry axis plane of the pier model along the centerline of the flume width.
[0064] Assuming different initial distances L between the central axis plane of the sand pit and the symmetry axis plane of the pier model k The curve in the figure is the L in the actual design. k Two-dimensional cross-sectional diagram of water flow conditions at 0.25m, 0.5m, 0.75m or 1m respectively.
[0065] Figure 6 In a, because the water velocity is greater than the sediment starting velocity, the sediment particles on the bed are in constant motion, causing the bed topography to fluctuate. U1 and a U2 They are basically equal to the sediment repose angle.
[0066] Figure 6 In b, due to the symmetrical flow structure on both sides of the pier model, the scour hole is symmetrical on the horizontal Y-Z plane, that is, the angle a between the elevation profile line and the horizontal Y axis is L =a R As the distance L between the sand pit and the bridge pier model k The closer the bridge pier is, the larger the radius of the scour pit becomes; the lower the bed terrain elevation is, that is, the deeper the riverbed is cut, the larger the exposed range of the bridge pier becomes.
[0067] The device can also be used to analyze the changes in pit shape at different distances between the initial sand pit centerline plane and the symmetry axis plane of the bridge pier model. The specific method includes quantitative calculation of the pit area at different depths for the scouring sand pit and the scouring pit of the bridge pier. The maximum pit depth is calculated by selecting the centerline plane of the sand pit and the symmetry axis plane of the bridge pier model and meshing the pit area. 坑 The N lowest elevation points Z processed i In this embodiment, N is set to 20.
[0068] The calculation of the pit area is based on the corresponding area of the maximum pit depth, the pit depth d s , pit area V 坑 The calculation formula is as follows.
[0069]
[0070] On the other hand, through Figure 7 The following is a two-dimensional cross-sectional view of the sand pit area formed after scouring at different distances between the sand pit and the bridge pier model obtained using this device:
[0071] The dark area (blue) is the cross section 102 of the scour topography map, and the light area (yellow) is the cross section 103 of the sand mining pit. Figure 7 It can be observed that, since the position of the pier model was fixed for each test, the position of the sand pit varied as the distance between it and the pier model varied. The X-axis range of the flow direction intercepted by the sand pit in the figure represents the initial range of the sand pit before scour. Horizontal observation reveals that as the pit depth increases, the cross-sectional area corresponding to the corresponding pit depth increases, i.e., the lighter-colored area becomes larger.
[0072] Vertical comparison shows that as the distance between the sand mining pit and the bridge pier model gets closer, the cross-sectional area of the sand mining pit at the same pit depth becomes larger.
[0073] like Figure 8 In the figure, the outer area is the bridge pier scour topographic map section 102 (blue), the central area circle is the bridge pier model section 101 (green), and the middle area enclosed by the bridge pier model section 101 is the bridge pier scour pit section 104 (yellow).
[0074] Horizontal observation reveals that as the sand pit deepens, when the cross-section of the pier scour pit does not completely enclose the pier model, the cross-section 104 of the pier scour pit increases symmetrically about the central axis of the pier model pit surface cross-section 104. However, when the cross-section 104 completely encloses the pier model, the cross-section 104 gradually evolves into an irregular heart shape. Vertical comparison clearly shows that as the distance between the sand pit and the pier model decreases, the cross-section area of the pier scour pit increases at the same depth, meaning the pit diameter also gradually increases.
[0075] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention's description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.
Claims
1. A reconstruction method for simulating local scour terrain of bridge piers in a sand pit environment, characterized by: The invention comprises a bridge pier local scour test model, which comprises a water tank, the bottom of which is paved with gravel, a vertical support rod simulating a bridge pier fixed in the middle of the water tank, and a concave sand pit area on the side of the vertical support rod; a circular track is arranged above the water tank, and a mobile camera that can move along the circular track is set on the circular track; one end of the water tank has a water inlet pipe and the other end has a drain outlet; Step (1): In the shooting area of the annular area to be tested, a black and white checkerboard is set as the terrain reconstruction control points, and the terrain reconstruction control points are evenly distributed; Step (2): Conduct a flume riverbed scour model test, release water from one side of the flume, and control the water flow rate and time according to the design requirements; then turn off the water flow, slowly reduce the flume flow to a stop, and use a mobile camera to take circular photos of the study area when there is no water accumulation in the study area; Step (3): The photos are input into the VisualSFM platform for calculation, and the coordinates of the set terrain reconstruction control points are input into the model. Feature points of multiple adjacent photos taken continuously are matched to identify the same feature points. On this basis, the PMVS algorithm is used to optimize the results of the sparse 3D point cloud reconstruction, that is, to generate a dense 3D point cloud of the scour bed surface. Step (4): Divide the sand-topped area into a grid of square areas, and use the interpolation method to process the point cloud data in the grid to obtain the interpolation results on the grid points; Step (5): Analyze the topographic characteristics of the local scour bed of the bridge pier when there is a sand pit and quantitatively calculate the pit area: Step (6): After the area to be measured is gridded, the maximum pit depth and pit area are calculated.
2. The reconstruction method for simulating local scour terrain of bridge piers in a sand pit environment according to claim 1, characterized in that: The calculation steps for quantitative calculation of the pit area in step (5) are as follows: first, subtract the slope elevation and level the calculation domain. The specific steps are as follows: Assume that the slope of the water tank is J = tanθ, there are n grids in the X direction, and the height of the slope at the i-th grid is h i =J(net n -net i ) Subtract the height of the tilt h from the height of the i-th grid i That is the terrain elevation when it is a horizontal plane, that is, the calculation domain is horizontal.
3. The reconstruction method for simulating local scour terrain of bridge piers in a sand pit environment according to claim 2, characterized in that: In step (6), the pit area of the bridge pier scour pit at different depths is quantitatively calculated. The maximum pit depth is calculated by selecting the plane near the central axis of the sand mining pit and the plane near the symmetry axis of the bridge pier model through the grid. 坑 The N lowest elevation points Z processed i The pit area is calculated based on the corresponding area of the maximum pit depth, the pit depth d s , pit area V 坑 The calculation formula is as follows:
4. The reconstruction method for simulating local scour terrain of bridge piers in a sand pit environment according to claim 2, characterized in that: The overlap rate of adjacent photos reaches more than 80%.
5. The reconstruction method for simulating local scour terrain of bridge piers in a sand pit environment according to claim 1, characterized in that: The annular track is fixed on the bracket, the bracket is fixed above the water tank, the vertical support rod is located in the middle of the water tank, the lower part of the vertical support rod is buried in the gravel layer, and the axial direction of the annular track above the water tank coincides with the axial direction of the vertical support rod.
6. The reconstruction method for simulating local scour terrain of bridge piers in a sand pit environment according to claim 5, characterized in that: The circular track is provided with a mobile vehicle that can move along the track. The mobile vehicle has a lower part with a circular track that cooperates with the circular track and an annular tooth surface on the outer side. The mobile vehicle is provided with a drive motor. The output end of the drive motor is fixed with a gear that meshes with the tooth surface. The drive motor is driven by the gear to move along the circular track. The mobile camera is fixed on the mobile vehicle.
7. The reconstruction method for simulating local scour terrain of bridge piers in a sand pit environment according to claim 1, characterized in that: In order to simulate the flow state of a river or simulate different water flow rates, the water tank has an inlet pipe at one end and a drain pipe at the other end; a flow valve is provided at the inlet pipe and a filter is provided at the drain pipe.