A multi-camera real-time monitoring method and system for the cantilever assembly posture of an arch bridge steel truss section

Abstract

The application discloses a kind of arch bridge steel truss section cantilever assembly posture multi-eye real-time monitoring method and system, belong to construction monitoring technical field, solve the problem that traditional precision level and total station rely on professional personnel operation and data processing, cannot monitor the dynamic posture change of cable crane tower and other structures under dynamic load, the method includes segment manufacturing form record, segment assembly posture photogrammetry, segment assembly posture is measured, based on segment assembly posture solution result real-time calculation current pose and design pose error, and complete segment main pipe splicing deviation correction;In the application, the spatial coordinates of all key points can be fed back in real time, and the posture of the hoisted segment is monitored in real time. Thus, accurate data support is provided for rapid correction. A 3-eye or more industrial camera array is used to ensure unobstructed imaging of at least three non-collinear targets under complex cantilever assembly conditions, significantly improving work efficiency and reducing construction cost.

Description

Technical Field

[0001] This invention belongs to the field of construction monitoring technology, specifically relating to a multi-view real-time monitoring method and system for the cantilever assembly posture of arch bridge steel truss segments. Background Technology

[0002] With the continuous advancement of transportation infrastructure construction in my country, large-span steel-concrete composite arch bridges have become a key bridge type for crossing rivers, lakes, and seas in complex terrains such as mountainous areas and canyons due to their strong spanning capacity, high structural rigidity, and excellent economic efficiency. The construction technology of such bridges directly affects project progress, structural safety, and the quality of the completed bridge. Among these methods, cable-stayed and cantilever assembly methods have become the mainstream construction methods for large-span steel-concrete composite arch bridges due to their high efficiency and reliability. The core of this method lies in achieving precise hoisting, temporary fixing, and alignment control of arch rib segments through the coordinated action of the cable towers, cable-stayed systems, and cables, ultimately completing the arch rib closure. However, as arch bridge construction expands into challenging mountainous areas, the construction environment becomes increasingly complex, span requirements continue to increase, labor costs rise year by year, and the demands for construction efficiency, safety, and the quality of the completed bridge are drastically increasing. The reliance of traditional construction techniques on digital and automated monitoring methods is becoming increasingly prominent. Among them, rapid and high-precision measurement technology is a key prerequisite for realizing real-time digital twins of arch bridges, dynamic control of the construction process, and high-quality construction. Its technical bottleneck has become the core issue restricting the improvement of the construction level of long-span arch bridges.

[0003] Currently, the industry mainly relies on the following measurement technologies for monitoring the spatial attitude and deformation of key structures (such as towers, cable-stayed systems, and arch rib segments) in the construction of long-span steel-concrete composite arch bridges: (1) Due to their high measurement accuracy (accuracy ±0.1mm or higher), bridge construction measurements often rely on instruments such as precision levels and total stations. This technology has the deepest application background, the widest application range, and the highest industry dependence. (2) Real-time Kinematic (RTK) is a GNSS real-time dynamic positioning technology (GPS, Beidou, etc.). This technology accurately determines the spatial relative position of the base station and the rover by receiving satellite signals between the base station and the rover and using carrier phase differential technology for differential calculation. RTK measurement results usually include the longitude, latitude, and elevation of the base station. In cable-stayed tower displacement measurement, these data can be converted into actual triaxial translational displacement values, which are applicable to various structures and have wide applicability. However, RTK technology is susceptible to GPS signal loss, which is particularly evident in complex mountainous bridge construction environments. It is often affected by obstructions from buildings, trees, and mountains, leading to a significant reduction in measurement accuracy. In addition, the use of RTK technology usually requires the establishment of base stations to provide reference signals, which increases the cost and complexity of the monitoring system. Secondly, GPS signal transmission distance is limited, usually only effective within a range of tens of kilometers, and exceeding this range will directly affect the measurement accuracy. Due to the high cost of GPS, monitoring systems are only set up in a few important bridge projects, which limits its applicability. (3) Machine vision measurement technology, with its advantages of non-contact, high precision, and real-time operation, is gradually being applied to bridge construction measurement. However, it is a relatively new technology in recent years, and its application is not yet widespread.

[0004] In summary, existing optical measurement methods relying on levels and total stations require specialized personnel for operation and data processing, resulting in high labor costs. The monitoring range is relatively limited, allowing only single-point measurements and is easily obstructed by line-of-sight, often necessitating station switching, leading to significant time consumption and accumulated errors. Secondly, this method can only be used for attitude deformation measurement under static loads, and cannot monitor attitude deformation of cable-stayed tower structures under dynamic loads. Thirdly, although the accuracy of RTK measurements generally meets practical engineering needs, its real-time data processing and feedback capabilities are poor, requiring substantial manpower and resources for post-processing, increasing construction costs; furthermore, since deformation is usually instantaneous, it is difficult to provide real-time feedback and timely warnings regarding attitude changes in cables, towers, arches, etc. Finally, existing bridge visual measurement methods generally employ monocular vision schemes, requiring the camera plane to be parallel to the target plane. However, this measurement condition is often difficult to meet in complex construction conditions in mountainous areas. Furthermore, technologies for long-distance visual measurement methods are even more lacking, significantly limiting their practicality. Furthermore, the existing spatial attitude deformation measurement technologies for towers, arches, and suspended objects of long-span bridges can only measure displacement values ​​and lack the ability to monitor the overall structure, thus failing to provide comprehensive and in-depth monitoring information. To address these issues, we propose a multi-view real-time monitoring method and system for the cantilever assembly attitude of arch bridge steel truss segments. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by providing a multi-view real-time monitoring method and system for the cantilever assembly posture of arch bridge steel truss segments. This solves the problems of traditional precision levels and total stations, which rely on professional personnel for operation and data processing, resulting in high labor costs. They also only support single-point measurements, are easily obstructed by mountains, towers, etc., and require frequent station changes, leading to wasted time and accumulated errors. Furthermore, they cannot monitor the dynamic posture changes of structures such as cable crane towers under dynamic loads.

[0006] This invention is implemented as follows: a multi-view real-time monitoring method for the cantilever assembly posture of arch bridge steel truss segments, comprising:

[0007] S10, Segment Manufacturing Morphology Recording: Calibrate four motion capture cameras, record the manufacturing morphology of the arch bridge steel truss segments based on the motion capture cameras, and retain the tube body target for subsequent measurements;

[0008] S20, Segment assembly posture photogrammetry: Pre-calibrate the intrinsic parameter matrix of the industrial RGB camera, perform photogrammetry on the segment assembly posture using the industrial RGB camera, calculate the segment assembly posture, and output the segment assembly posture calculation results;

[0009] S30, Segment Master Splicing Correction: Load at least one set of segment assembly attitude calculation results, calculate the error between the current pose and the design pose in real time based on the segment assembly attitude calculation results, determine whether the error between the overall current pose and the design pose of the segment meets the preset error threshold, and complete the segment master splicing correction.

[0010] Preferably, the method for recording the manufacturing morphology of arch bridge steel truss segments based on motion capture cameras includes:

[0011] S101, calibration of four motion capture cameras in the assembly field: point cloud scale is calculated using a fixed-length calibration rod. During point cloud scale calculation, reflector ball holders are fixedly installed at both ends of the calibration rod. The reflector ball holder consists of a reflector ball, a magnetic base, and a screw. The reflector ball is detachably installed in the magnetic base via the screw. The distance between the two sets of reflector balls is 1 meter.

[0012] S102 uses a motion capture camera to capture and process images of reflective spots, captures images of reflective spheres and reflective targets, and uses the RANSAC algorithm to obtain and number the pixel coordinates of the image center point.

[0013] S103, using multi-view geometry or 3D reconstruction algorithms to calculate the centroid world coordinates of the target and the reflective sphere, and to calculate the world coordinates of the reflective sphere and the reflective target with the pixel coordinates of the center point of the image with different numbers.

[0014] S104: Fit the center coordinates of the end of the main pipe segment and convert them into the local coordinates of the segment. Repeat steps S101-S104 for the pipe segments being cut. Then remove the reflective ball fixer and retain the pipe body target for attitude measurement and calculation during subsequent segment assembly.

[0015] Preferably, the method for fitting the coordinates of the center of the end of the main pipe segment and converting them into local coordinates of the segment includes:

[0016] S1041, triangulate the pixels in the imaging plane of the four motion capture cameras, and obtain the world coordinates of the center point of the i-th marker point through bundle adjustment optimization of multiple views;

[0017] S1042, According to the least squares fitting, obtain the spatial center coordinates of each main pipe end in the segment, wherein the spatial center of the steel pipe end and the center of the circle formed by several reflective balls fixed on the end flange are on the same axis.

[0018] S1043, map the centroid coordinates of the end flange of the segment into the local coordinate system of the segment, and transform the world coordinates of the center of the two ends of the segment into the local coordinate system of the pipe body.

[0019] Preferably, the method for calculating the segment assembly attitude includes:

[0020] S201 Before installing the first section, set up no less than 3 non-collinear permanent control points within the field of view of each industrial RGB camera on the opposite bank, and use a total station to measure and obtain the world coordinates of the control points, and calibrate the intrinsic parameters of the industrial RGB cameras set on the sides of the arch feet on both banks.

[0021] S202, to take long-distance photos of the target segments and calculate the world coordinates of the target;

[0022] S203 obtains the pose information of the current node through inverse local coordinate transformation, and transforms the center coordinates of all flange holes at the end of the segment based on the pose information of the current node, realizing the inverse transformation from local coordinates to world coordinates.

[0023] Preferably, the method for correcting the splicing of the main segment includes:

[0024] S301, Load at least one set of segment assembly attitude calculation results and obtain segment design pose, and calculate the overall pose of the segment;

[0025] S302, adjust the overall pose of the segment based on the design pose, calculate the error between the current pose and the design pose in real time based on the segment assembly pose calculation result, output the overall pose calculation of the segment, and determine whether the error between the current pose and the design pose of the segment meets the preset error threshold.

[0026] S303, if the error between the current pose of the entire segment and the design pose meets the preset error threshold, the flanges of adjacent segments are tightened.

[0027] S304, interlocking and tensioning of the cable;

[0028] S305, main pipe closed circumferential welding, completes the construction of this section.

[0029] Preferably, before calculating the segment assembly attitude, the intrinsic parameter matrix of four industrial RGB cameras is pre-calibrated. In subsequent photogrammetry operations, three or more industrial RGB cameras are aligned with the four targets pasted on the bottom of the assembled segment to be tested, and no fewer than two industrial RGB cameras are used to image the targets completely.

[0030] Preferably, the industrial RGB camera consists of two medium-high resolution cameras and two high resolution cameras, with one set on each side of the arch bridge. The high resolution camera is used to observe the target on the opposite bank segment, while the medium-high resolution camera is used to observe the target on the local bank. On arch bridges with a span of more than 300m, the medium-high resolution camera has more than 70 million pixels, and the target size is greater than 30cm. The target is made of reflective material.

[0031] Preferably, in step S303, if the error between the current pose of the segment as a whole and the design pose does not meet the preset error threshold, the segment main pipe splicing is corrected based on the design pose, the construction equipment is triggered to adjust the parameters of the industrial RGB camera and steps S201-S203 are re-executed until the error between the current pose of the segment as a whole and the design pose meets the preset error threshold.

[0032] On the other hand, the present invention also provides a multi-view real-time monitoring system for the cantilever assembly attitude of arch bridge steel truss segments, the multi-view real-time monitoring system for the cantilever assembly attitude of arch bridge steel truss segments includes:

[0033] The segment manufacturing morphology recording module is used to calibrate four motion capture cameras, record the manufacturing morphology of the arch bridge steel truss segments based on the motion capture cameras, and retain the tube body target for subsequent measurements.

[0034] The segment assembly attitude measurement module is used to pre-calibrate the intrinsic parameter matrix of the industrial RGB camera, perform video measurement of the segment assembly attitude through the industrial RGB camera, calculate the segment assembly attitude, and output the segment assembly attitude calculation result.

[0035] The main splicing correction module is used to load at least one set of segment assembly attitude calculation results, calculate the error between the current pose and the design pose in real time based on the segment assembly attitude calculation results, determine whether the error between the overall current pose and the design pose of the segment meets the preset error threshold, and complete the segment main splicing correction.

[0036] Preferably, the main splicing and correction module includes:

[0037] The overall pose calculation unit is used to load at least one set of segment assembly attitude calculation results and integrate the segment assembly attitude calculation results to obtain the overall pose calculation result of the segment.

[0038] The error calculation unit calculates the error between the current pose and the designed pose in real time based on the segment assembly attitude calculation results;

[0039] The error judgment unit is used to determine whether the error between the current pose of the whole segment and the designed pose meets the preset error threshold.

[0040] The correction control unit triggers a camera parameter adjustment strategy based on the error between the current pose and the designed pose of the entire segment, and completes the segment main splicing correction based on the camera parameter adjustment strategy.

[0041] Compared with the prior art, the embodiments of this application have the following main advantages:

[0042] In this embodiment of the invention, the spatial coordinates of all key points can be fed back in real time, enabling real-time monitoring of the attitude of the hoisting segments. This provides accurate data support for rapid correction. A three- or more industrial camera array is employed, with high- and low-resolution cameras on both sides, ensuring unobstructed imaging of at least three non-collinear targets under complex cantilever assembly conditions. This solves the problem of station relocation caused by single-view line-of-sight obstruction and avoids the drawbacks of traditional measurements where work is forced to be interrupted due to obstruction. Furthermore, by using a priori storage of segment geometry, only three targets on the main pipe need to be measured during the formal segment splicing stage. The centroid coordinates at both ends of the main pipe are calculated using the target coordinates, greatly reducing the number of measurement points, significantly improving work efficiency, and lowering construction costs. Attached Figure Description

[0043] Figure 1 This is a schematic diagram illustrating the implementation process of the multi-view real-time monitoring method for the cantilever assembly posture of arch bridge steel truss segments provided by the present invention.

[0044] Figure 2 The diagram shows the shape of the steel pipe segment and the target measurement.

[0045] Figure 3 The diagram shows the spatial arrangement of industrial RGB cameras during the assembly of the main arch segments of the arch bridge.

[0046] Figure 4 The diagram shows the plan and elevation layout of the industrial RGB camera when the main arch segment of the arch bridge is being assembled.

[0047] Figure 5 A schematic diagram of the gimbal rotation angle of an industrial RGB camera is shown.

[0048] Figure 6 This diagram illustrates the imaging and target key point matching process when an industrial RGB camera performs long-distance shooting of a segmented target.

[0049] Figure 7 This is a structural schematic diagram of the multi-view real-time monitoring system for the cantilever assembly posture of arch bridge steel truss segments provided by the present invention.

[0050] In the diagram: 1. Segment; 2. Target; 3. Reflector ball holder; 3-1. Reflector ball; 3-2. Screw; 3-3. Magnetic base; 4. Motion capture camera; 4-1. Imaging plane; 4-2. Center pixel; 5. Industrial RGB camera; 5-1. Medium-to-high resolution camera; 5-2. High resolution camera; 6. Flange;

[0051] 100. Segment manufacturing form recording module; 200. Segment assembly posture measurement module; 300. Main pipe splicing correction module; 310. Overall posture calculation unit; 320. Error calculation unit; 330. Error judgment unit; 340. Correction control unit. Detailed Implementation

[0052] 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 belongs; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and foregoing drawings of this application are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification, claims, or foregoing drawings of this application are used to distinguish different objects, not to describe a particular order.

[0053] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0054] Traditional precision levels and total stations rely on professional operators and data processing, resulting in high labor costs. They also only support single-point measurements, are easily obstructed by mountains, towers, etc., requiring frequent station changes, leading to wasted time and accumulated errors. Furthermore, they cannot monitor the dynamic attitude changes of structures such as cable cranes under dynamic loads. To address these issues, we propose a multi-view real-time monitoring method and system for the cantilever assembly attitude of arch bridge steel truss segments. In short, the method achieves multi-view real-time monitoring of the cantilever assembly attitude of arch bridge steel truss segments through segment manufacturing morphology recording, segment assembly attitude photogrammetry, and segment main pipe splicing correction. In this embodiment, the spatial coordinates of all key points can be fed back in real time, enabling real-time monitoring of the attitude of the hoisted segment 1. This provides accurate data support for rapid deviation correction. By using an array of 3 or more industrial cameras and combining high and low resolution cameras on both sides, it ensures unobstructed imaging of at least 3 non-collinear targets 2 under complex cantilever assembly conditions. This solves the problem of station relocation caused by single-view line-of-sight obstruction and avoids the drawback of forced work interruption due to obstruction in traditional measurement. In addition, by using the method of storing the geometry of segment 1 in advance, only 3 targets 2 on the main pipe need to be measured during the formal segment splicing stage. The coordinates of the targets 2 are used to calculate the centroid coordinates of the two ends of the main pipe, which greatly reduces the number of measurement points, significantly improves work efficiency, and reduces construction costs.

[0055] This invention provides a multi-view real-time monitoring method for the cantilever assembly posture of arch bridge steel truss segments. Figure 1A schematic diagram illustrating the implementation process of a multi-view real-time monitoring method for the cantilever assembly attitude of arch bridge steel truss segments is shown. The method specifically includes:

[0056] S10, segment manufacturing morphology recording: calibrate four motion capture cameras 4, record the manufacturing morphology of arch bridge steel truss segment 1 based on motion capture cameras 4, and retain the tube body target 2 for subsequent measurements.

[0057] S20, Segment assembly posture photogrammetry: Pre-calibrate the intrinsic parameter matrix of industrial RGB camera 5, perform photogrammetry on the assembly posture of segment 1 through industrial RGB camera 5, calculate the assembly posture of segment 1, and output the assembly posture calculation result of segment 1.

[0058] S30, Segment Master Splicing Correction: Load at least one set of segment 1 assembly attitude calculation results, calculate the error between the current pose and the design pose in real time based on the segment 1 assembly attitude calculation results, determine whether the error between the overall current pose and the design pose of segment 1 meets the preset error threshold, and complete the segment master splicing correction.

[0059] In this embodiment of the invention, the spatial coordinates of all key points can be fed back in real time, enabling real-time monitoring of the attitude of the hoisting segment 1. This provides accurate data support for rapid correction. A three- or more industrial camera array is used, with high- and low-resolution cameras on both sides, ensuring unobstructed imaging of at least three non-collinear targets 2 under complex cantilever assembly conditions. This solves the problem of station relocation caused by single-view line-of-sight obstruction and avoids the drawbacks of traditional measurements where work is forced to be interrupted due to obstruction. Furthermore, by using a priori storage of the segment 1's geometry, only three targets 2 on the main pipe need to be measured during the formal segment splicing stage. The centroid coordinates of both ends of the main pipe are calculated using the target 2 coordinates, greatly reducing the number of measurement points, significantly improving work efficiency, and lowering construction costs.

[0060] In this embodiment of the invention, the method for recording the manufacturing morphology of the arch bridge steel truss segment 1 based on the motion capture camera 4 includes:

[0061] S101, four motion capture cameras in the assembly field are calibrated: the point cloud scale is calculated using a fixed-length calibration rod.

[0062] It should be noted that segment 1 refers to several standardized and modular prefabricated component units that the overall bridge structure (such as the main arch, beams, trusses, etc.) is pre-divided according to design requirements. During point cloud scale calculation, reflective ball holders 3 are fixedly installed at both ends of the calibration rod. The reflective ball holder 3 consists of a reflective ball 3-1, a magnetic base 3-3, and a screw 3-2. The reflective ball 3-1 is detachably installed in the magnetic base 3-3 via the screw 3-2. The distance between the two sets of reflective balls 3-1 is 1 meter. For example... Figure 2The diagram shows the shape of the steel pipe segment 1 and the target measurement schematic diagram, that is, the schematic diagram of the reflective ball fixing device 3 installed at both ends of the steel pipe segment 1. The reflective ball fixing device 3 is detachably installed on the flange 6 through the magnetic base 3-3. The flange 6 is embedded in the steel pipe segment 1. In this embodiment, the reflective ball 3-1 is a marker ball.

[0063] S102, uses motion capture camera 4 to capture and process images of reflective spots, captures images of reflective sphere 3-1 and reflective target 2, and uses RANSAC algorithm to obtain the coordinates of the center point pixel 4-2 of the image and number it;

[0064] In this embodiment of the invention, when capturing images of the reflective sphere 3-1 and the reflective target 2, grayscale images are captured and processed. The RANSAC algorithm is used to obtain the coordinates of the center pixel 4-2 of the images of the reflective sphere 3-1 and the reflective target 2 captured by each motion capture camera 4, and these center pixel 4-2s are numbered to facilitate subsequent triangulation calculations. The centerline of the reflective sphere holder 3 should be aligned with the axis of symmetry of the flange 6 to accurately determine the position and orientation of the flange 6.

[0065] S103, use multi-view geometry or three-dimensional reconstruction algorithm to solve the centroid world coordinates of target 2 and reflective sphere 3-1, and solve the world coordinates of reflective sphere 3-1 and reflective target 2 with the coordinates of pixel 4-2, the center point of images with different numbers;

[0066] In this embodiment of the invention, when using multi-view geometry or three-dimensional reconstruction algorithms to calculate the centroid world coordinates of target 2 and reflective sphere 3-1, OpenMVG or Colmap algorithms can be used, or the projection matrix method of formula (10) in the subsequent steps can be used to calculate the world coordinates of reflective sphere 3-1 and reflective target 2 with different numbers.

[0067] S104, Fit the center coordinates of the end of the main pipe of segment 1 and convert them into the local coordinates of segment 1. Repeat steps S101-S104 for the pipe segments to be cut. Then remove the reflective ball fixer 3 and retain the pipe body target 2 for attitude measurement and calculation during subsequent segment 1 assembly.

[0068] In this embodiment of the invention, the method for fitting the coordinates of the center of the end of the main tube of the segment and converting them into local coordinates of the segment includes:

[0069] S1041, triangulate the pixels in the imaging plane 4-1 of the four motion capture cameras 4, and obtain the world coordinates of the center point of the i-th marker point through bundle adjustment (BA) of multiple views;

[0070] It should be noted that in this embodiment, the i-th marker point is the reflective ball 3-1 and the reflective target 2.

[0071] S1042, based on least squares fitting, obtain the spatial center coordinates of the end of each main pipe in segment 1. The spatial center of the end of the steel pipe lies on the same axis as the center of the circle formed by the several reflective spheres 3-1 fixed on the end flange 6. Therefore, the world coordinates of the end center of the steel pipe can be fitted using the spatial point of the center of the reflective spheres 3-1. The least squares expression is as follows:

[0072]

[0073] Among them, XO i,j ,YO i,j ZO i,j Let be the world coordinates of the center of the j-th end opening of the i-th steel pipe segment; R represents the set of X, Y, Z coordinates of the center point of the reflective sphere 3-1 at the j-th end of the i-th steel pipe segment in the local coordinate system. i,j This represents the outer radius of the i-th arch rib segment 1 and the j-th flange 6.

[0074] S1043, map the centroid coordinates of the flange 6 at the end of segment 1 into the local coordinate system of segment 1, and transform the world coordinates of the centers at both ends of segment 1 into the local coordinate system of the pipe body.

[0075] It should be noted that mapping the centroid coordinates of the end flange 6 of segment 1 into the local coordinate system of segment 1 is called local coordinate transformation. First, any three reflective targets 2 are selected as reference points for the local coordinate system, and the local coordinate system Coor of the i-th segment 1 is established. As in formula (2-4) and Figure 2 As shown, where norm represents normalization:

[0076]

[0077] in, Let x, y, and z represent the world coordinates of the center point of the j-th reflective target 2, respectively. A vector is obtained by the cross product of two vectors.

[0078] In this embodiment of the invention, the world coordinates of the centers at both ends of segment 1 are transformed to the local coordinate system of the pipe body, as shown in formula (5):

[0079]

[0080] in,

[0081] Where lowercase x, y, and z represent world coordinates, uppercase X, Y, and Z represent local coordinates, and T represents world coordinates. i Let represent the local coordinate system transformation matrix of the target 2 in the i-th segment 1.

[0082] It should be noted that in step S104, since the horizontally assembled segment 1 is placed on the ground, the target 2 should be attached to the belly or bottom surface of segment 1 to avoid subsequent imaging occlusion. The number of targets 2 on each segment 1 should not be less than 3 and they should not be collinear. It is advisable to set 4 at the bottom, roughly forming a rectangle, and the larger the enclosed area, the better, to avoid scene degradation due to occlusion or loss of pose information due to collinearity ambiguity during calculation, ensuring the robustness of recording the relative spatial relationship of the steel pipe ports. Repeat steps S101-S104 for each steel pipe segment, then remove the reflective ball holder 3, retaining the pipe body target 2 for attitude measurement and calculation during subsequent segment 1 assembly.

[0083] In this embodiment of the invention, before calculating the assembly posture of segment 1, the intrinsic parameter matrices of four industrial RGB cameras 5 are pre-calibrated. In subsequent photogrammetry operations, it is ensured that three or more industrial RGB cameras 5 are aligned with the four targets 2 pasted on the bottom of segment 1 in the assembly test state, and at least two industrial RGB cameras 5 completely image the targets 2. The optimal effect is that all four cameras simultaneously and completely image the targets.

[0084] To reduce equipment costs, the industrial RGB camera 5 can be divided into two medium-to-high resolution cameras 5-1 and two high resolution cameras 5-2, with one set deployed on each side of the arch bridge. The high resolution camera 5-2 is used to observe the target 2 on the opposite bank segment 1, while the medium-to-high resolution camera 5-1 is used to observe the local bank. Figure 3 The diagram shows the spatial arrangement of the industrial RGB camera 5 during the assembly of the main arch segment 1 of the arch bridge. Figure 4 The diagram shows the plan and elevation layout of the industrial RGB camera 5 when the main arch segment 1 of the arch bridge is assembled. In order to ensure that the measurement accuracy reaches the centimeter level, on arch bridges with a span of more than 300m, the medium and high resolution camera 5-1 has a pixel size of more than 70 million pixels, and the target 2 is larger than 30cm in size. The target 2 is made of reflective material to facilitate image processing and feature detection.

[0085] In a further preferred embodiment of the present invention, a method for calculating the attitude of segment assembly is provided, the method specifically including:

[0086] S201 Before installing the first section 1, set up no less than 3 non-collinear permanent control points within the field of view of each industrial RGB camera 5 on the opposite bank, and use a total station to measure and obtain the world coordinates of the control points, and calibrate the intrinsic parameters of the industrial RGB cameras 5 set on the sides of the arch feet on both banks.

[0087] In step S201, at least three non-collinear permanent control points are first set within the field of view of each industrial RGB camera 5 on the opposite bank. The world coordinates of these control points are obtained beforehand using a total station. Then, during the initial calibration of the industrial RGB cameras 5, the extrinsic parameter matrix of each camera 5 is calibrated using the PnP method, based on the permanent control points within its field of view. This matrix is ​​{R|T}, where R is the rotation matrix of the camera attitude and T represents the translation vector. The intrinsic parameter matrix K should be calibrated before the industrial RGB cameras 5 are installed. If the angle of the industrial RGB cameras 5 is adjusted during the installation and measurement of subsequent segments 1, the world coordinates of the target 2 in the previous segment 1 are used as control points for PnP extrinsic parameter calibration. Thus, at each construction stage, the world coordinates of the measuring points can be obtained from the projection matrix.

[0088] S202, perform long-distance imaging of target 2 in segment 1, and calculate the world coordinates of target 2. Figure 5 A schematic diagram of the gimbal rotation angle of the industrial RGB camera 5 is shown, while Figure 6 The diagram shows the imaging and key point matching of the target 2 when the industrial RGB camera 5 takes a long-distance picture of the target 2 in segment 1;

[0089] In this embodiment of the invention, when calculating the world coordinates (x, y, z) of target 2, the extrinsic parameter {R|T} is obtained using the known world coordinates (x, y, z) of the control point (key point of target 2) as described above. Since the intrinsic parameter K is known, the projection matrix of the industrial RGB camera 5 can be obtained by formula (7):

[0090] P = K·[R|T](7)

[0091] The control point and all targets 2 of the current segment 1 can be observed simultaneously in the image. The pixel coordinates of the center point of these targets 2 in the i-th industrial RGB camera 5 image are (u i v i For each industrial RGB camera 5, the pixel coordinates (u) i v i According to the projection relationship:

[0092]

[0093] After eliminating the scale factor s i Then, the system of linear equations corresponding to 5 frames from a single industrial RGB camera was obtained:

[0094]

[0095] The linear equations (9) from three industrial RGB cameras 5 are combined to obtain an overdetermined equation system (10). If four industrial RGB cameras 5 are used for solving, the number of elements will increase accordingly.

[0096]

[0097] in: Represents the projection matrix P i Given the element in the m-th row and n-th column, rewrite the system of equations (10) in the form Ax = b, then:

[0098]

[0099] Where i is the number of the industrial RGB camera 5, and its maximum value is the number of industrial RGB cameras 5 selected. For example, if the images from four industrial RGB cameras 5 are used to solve the problem, the number of rows in A and b above will be expanded to 8, and the solution can be obtained through linear least squares. Obtain the world coordinates (x, y, z).

[0100] S203, by inverse transformation of local coordinates, obtain the pose information of the current node, and transform the center coordinates of all flange holes at the end of segment 1 based on the pose information of the current node, thereby realizing the inverse transformation of local coordinates to world coordinates.

[0101] In this embodiment, as segment 1 undergoes continuous pose changes during assembly, its local coordinate system also continuously changes. To perform the inverse transformation, the coordinate axis vectors of segment 1's local coordinate system need to be reconstructed in real time. Therefore, formulas (2-4) need to be repeatedly calculated for each frame of the image. Based on the world coordinates of target 2 corresponding to each frame obtained in step S202, the local coordinate system Coor described in step S102 is continuously refreshed. and coordinate transformation matrix T i Therefore, the world coordinates (x, y, z) of the center of each steel pipe port and the center of the flange hole on the current segment 1 can be continuously refreshed and calculated using formula (11).

[0102]

[0103] In this context, lowercase x, y, and z represent world coordinates, while uppercase X, Y, and Z represent local coordinates. The inverse transformation matrix is ​​obtained by using formula (6).

[0104] In a further preferred embodiment of the present invention, the method for correcting segmental main pipe splicing includes:

[0105] S301, load at least one set of segment 1 assembly attitude calculation results and obtain segment design pose, and calculate the overall pose of segment 1;

[0106] S302, adjust the overall pose of segment 1 based on the design pose, calculate the error between the current pose and the design pose in real time based on the assembly pose calculation result of segment 1, output the overall pose calculation of segment 1, and determine whether the error between the current pose and the design pose of segment 1 meets the preset error threshold.

[0107] It should be noted that, compared with the designed pose, the error between the current pose and the designed pose includes the spatial pose error of segment 1 in the current state and the specific point deviation. The overall error between the current pose and the designed pose of segment 1 is calculated using formula 12:

[0108]

[0109] Where Δ i, j is the deviation of the center of the j-th pipe opening on the i-th steel pipe segment 1; These are the current coordinates of the j-th nozzle on the i-th segment 1; These are the design coordinates of the j-th nozzle on the i-th segment 1.

[0110] S303, if the error between the current pose of segment 1 and the design pose meets the preset error threshold, the flange 6 of the adjacent segment 1 is tightened; then the hoisting work of the next segment 1 is carried out, the gear is switched, and the steps S201-S203 are returned to adjust the camera gimbal angle, focus, and recalibrate.

[0111] It should be noted that in step S303, if the error between the current pose of segment 1 and the design pose does not meet the preset error threshold, the segment main body splicing correction is performed based on the design pose, triggering the construction equipment to adjust the parameters of the industrial RGB camera 5 and re-execute steps S201-S203 until the error between the current pose of the segment and the design pose meets the preset error threshold.

[0112] S304, interlocking and tensioning of the cable;

[0113] S305, main pipe closed ring welding, complete the construction of section 1.

[0114] In this embodiment of the invention, by solidifying the spatial geometric relationship between the center of the main pipe of segment 1 and the flange 6 during the manufacturing stage, there is no need to repeatedly measure redundant measuring points during assembly. The attitude of the entire segment 1 can be reconstructed based on only 3 reference targets 2, upgrading the traditional "point-by-point measurement-point-by-point calculation" mode to "prior model-incremental update", reducing the measurement time of a single segment 1.

[0115] On the other hand, the present invention also provides a multi-view real-time monitoring system for the cantilever assembly posture of arch bridge steel truss segments. Figure 7This diagram illustrates the structure of a multi-view real-time monitoring system for the cantilever assembly attitude of arch bridge steel truss segments. The system specifically includes:

[0116] The segment manufacturing morphology recording module 100 is used to calibrate four motion capture cameras 4, record the manufacturing morphology of the arch bridge steel truss segment 1 based on the motion capture cameras 4, and retain the tube body target 2 for subsequent measurements.

[0117] The segment assembly attitude measurement module 200 is used to pre-calibrate the intrinsic parameter matrix of the industrial RGB camera 5, perform video measurement on the assembly attitude of segment 1 through the industrial RGB camera 5, calculate the assembly attitude of segment 1, and output the assembly attitude calculation result of segment 1.

[0118] The main splicing correction module 300 is used to load at least one set of segment 1 assembly attitude calculation results, calculate the error between the current pose and the design pose in real time based on the segment 1 assembly attitude calculation results, determine whether the error between the overall current pose and the design pose of segment 1 meets the preset error threshold, and complete the segment main splicing correction.

[0119] In this embodiment, the main splicing and correction module 300 includes:

[0120] The overall pose calculation unit 310 is used to load at least one set of segment 1 assembly pose calculation results and integrate the segment 1 assembly pose calculation results to obtain the overall pose calculation result of segment 1.

[0121] Error calculation unit 320 calculates the error between the current pose and the designed pose in real time based on the pose calculation results of segment 1 assembly;

[0122] Error judgment unit 330 is used to judge whether the error between the current pose of segment 1 and the designed pose meets the preset error threshold.

[0123] The correction control unit 340 triggers a camera parameter adjustment strategy based on the error between the current pose and the designed pose of segment 1, and completes segment main body stitching correction based on the camera parameter adjustment strategy.

[0124] It should be noted that the multi-view real-time monitoring system for the cantilever assembly posture of arch bridge steel truss segments provided in this embodiment of the invention corresponds to the steps of the multi-view real-time monitoring method for the cantilever assembly posture of arch bridge steel truss segments described above. For details, please refer to the implementation steps of the multi-view real-time monitoring method for the cantilever assembly posture of arch bridge steel truss segments described above, which will not be repeated here.

[0125] In summary, this invention provides a multi-view real-time monitoring method and system for the attitude of cantilevered arch bridge steel truss segments. In this embodiment, the spatial coordinates of all key points can be fed back in real time, enabling real-time monitoring of the attitude of the hoisted segment 1. This provides accurate data support for rapid deviation correction. Using a three-view or higher industrial camera array, and combining high- and low-resolution cameras on both banks, it ensures unobstructed imaging of at least three non-collinear targets 2 under complex cantilever assembly conditions. This solves the problem of station relocation caused by single-view line-of-sight obstruction and avoids the drawbacks of traditional measurements where work is forced to be interrupted due to obstruction. Furthermore, by employing a priori storage of the segment 1's geometry, only three targets 2 on the main pipe need to be measured during the formal segment splicing stage. The centroid coordinates at both ends of the main pipe are calculated using the target 2 coordinates, greatly reducing the number of measurement points, significantly improving work efficiency, and lowering construction costs.

[0126] It should be noted that, for the sake of simplicity, the foregoing embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to the present invention. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0127] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit the scope of protection of the invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on these embodiments, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art can still combine, add, delete, or otherwise adjust the features of the various embodiments of the present invention according to the circumstances without conflict or creative effort, thereby obtaining different technical solutions that do not fundamentally depart from the concept of the present invention. These technical solutions also fall within the scope of protection of the present invention.

Claims

1. A method for real-time multi-view monitoring of the cantilever assembly posture of arch bridge steel truss segments, characterized in that, The method includes: S10, Segment Manufacturing Morphology Recording: Calibrate four motion capture cameras, record the manufacturing morphology of the arch bridge steel truss segments based on the motion capture cameras, and retain the tube body target for subsequent measurements; S20, Segment assembly posture photogrammetry: Pre-calibrate the intrinsic parameter matrix of the industrial RGB camera, perform photogrammetry on the segment assembly posture using the industrial RGB camera, calculate the segment assembly posture, and output the segment assembly posture calculation results; S30, Segment main body splicing correction: Load at least one set of segment assembly attitude calculation results, calculate the error between the current pose and the design pose in real time based on the segment assembly attitude calculation results, determine whether the error between the overall current pose and the design pose of the segment meets the preset error threshold, and complete the segment main body splicing correction. The method for recording the manufacturing morphology of arch bridge steel truss segments based on motion capture cameras includes: S101, calibration of four motion capture cameras in the assembly field: point cloud scale is calculated using a fixed-length calibration rod. During point cloud scale calculation, reflector ball holders are fixedly installed at both ends of the calibration rod. The reflector ball holder consists of a reflector ball, a magnetic base, and a screw. The reflector ball is detachably installed in the magnetic base via the screw. The distance between the two sets of reflector balls is 1 meter. S102 uses a motion capture camera to capture and process images of reflective spots, captures images of reflective spheres and reflective targets, and uses the RANSAC algorithm to obtain and number the pixel coordinates of the image center point. S103, using multi-view geometry or 3D reconstruction algorithms to calculate the centroid world coordinates of the target and the reflective sphere, and to calculate the world coordinates of the reflective sphere and the reflective target with the pixel coordinates of the center point of the image with different numbers. S104, Fit the center coordinates of the end of the main pipe segment and convert them into the local coordinates of the segment. Repeat steps S101-S104 for the pipe segments to be cut. Then remove the reflective ball fixer and retain the pipe body target for attitude measurement and calculation during subsequent segment assembly. The method for calculating the segment assembly attitude includes: S201 Before installing the first section, set up no less than 3 non-collinear permanent control points within the field of view of each industrial RGB camera on the opposite bank, and use a total station to measure and obtain the world coordinates of the control points, and calibrate the intrinsic parameters of the industrial RGB cameras set on the sides of the arch feet on both banks. S202, to take long-distance photos of the target segments and calculate the world coordinates of the target; S203, by inverse transformation of local coordinates, obtains the pose information of the current node, and transforms the center coordinates of all flange holes at the end of the segment based on the pose information of the current node, realizing the inverse transformation of local coordinates to world coordinates; The method for correcting the splicing of the main segment includes: S301, Load at least one set of segment assembly attitude calculation results and obtain segment design pose, and calculate the overall pose of the segment; S302, adjust the overall pose of the segment based on the design pose, calculate the error between the current pose and the design pose in real time based on the segment assembly pose calculation result, output the overall pose calculation of the segment, and determine whether the error between the current pose and the design pose of the segment meets the preset error threshold. S303, if the error between the current pose of the entire segment and the design pose meets the preset error threshold, the flanges of adjacent segments are tightened. S304, interlocking and tensioning of the cable; S305, main pipe closed circumferential welding, completes the construction of this section.

2. The method for multi-view real-time monitoring of the cantilever assembly posture of arch bridge steel truss segments as described in claim 1, characterized in that: The method for fitting the coordinates of the center of the end of the main tube of the segment and converting them into local coordinates of the segment includes: S1041, triangulate the pixels in the imaging plane of the four motion capture cameras, and obtain the world coordinates of the center point of the i-th marker point through bundle adjustment optimization of multiple views; S1042, According to the least squares fitting, obtain the spatial center coordinates of each main pipe end in the segment, wherein the spatial center of the steel pipe end and the center of the circle formed by several reflective balls fixed on the end flange are on the same axis. S1043, map the centroid coordinates of the end flange of the segment into the local coordinate system of the segment, and transform the world coordinates of the center of the two ends of the segment into the local coordinate system of the pipe body.

3. The method for multi-view real-time monitoring of the cantilever assembly posture of arch bridge steel truss segments as described in claim 2, characterized in that: Before calculating the assembly posture of the segments, the intrinsic parameter matrix of the four industrial RGB cameras is pre-calibrated. In subsequent photogrammetry operations, ensure that three or more industrial RGB cameras are aligned with the four targets pasted on the bottom of the assembled segments, and that at least two industrial RGB cameras are used to image the targets completely.

4. The method for multi-view real-time monitoring of the cantilever assembly posture of arch bridge steel truss segments as described in claim 3, characterized in that: The industrial RGB camera consists of two medium-high resolution cameras and two high resolution cameras, with one set on each side of the arch bridge. The high resolution camera is used to observe the target on the opposite bank segment, while the medium-high resolution camera is used to observe the target on the local bank. On arch bridges with a span of more than 300m, the medium-high resolution camera has more than 70 million pixels, and the target size is greater than 30cm. The target is made of reflective material.

5. The method for multi-view real-time monitoring of the cantilever assembly posture of arch bridge steel truss segments as described in claim 4, characterized in that: In step S303, if the error between the current pose of the segment and the design pose does not meet the preset error threshold, the segment main pipe splicing is corrected based on the design pose, the construction equipment is triggered to adjust the parameters of the industrial RGB camera and steps S201-S203 are re-executed until the error between the current pose of the segment and the design pose meets the preset error threshold.

6. A multi-view real-time monitoring system for the cantilever assembly posture of arch bridge steel truss segments, implemented using the multi-view real-time monitoring method for the cantilever assembly posture of arch bridge steel truss segments as described in claim 1, characterized in that: The multi-view real-time monitoring system for the cantilever assembly attitude of the arch bridge steel truss segments includes: The segment manufacturing morphology recording module is used to calibrate four motion capture cameras, record the manufacturing morphology of the arch bridge steel truss segments based on the motion capture cameras, and retain the tube body target for subsequent measurements. The segment assembly attitude measurement module is used to pre-calibrate the intrinsic parameter matrix of the industrial RGB camera, perform video measurement of the segment assembly attitude through the industrial RGB camera, calculate the segment assembly attitude, and output the segment assembly attitude calculation result. The main splicing correction module is used to load at least one set of segment assembly attitude calculation results, calculate the error between the current pose and the design pose in real time based on the segment assembly attitude calculation results, determine whether the error between the overall current pose and the design pose of the segment meets the preset error threshold, and complete the segment main splicing correction.

7. The multi-view real-time monitoring system for the cantilever assembly posture of arch bridge steel truss segments as described in claim 6, characterized in that: The supervisor splicing and correction module includes: The overall pose calculation unit is used to load at least one set of segment assembly attitude calculation results and integrate the segment assembly attitude calculation results to obtain the overall pose calculation result of the segment. The error calculation unit calculates the error between the current pose and the designed pose in real time based on the segment assembly attitude calculation results; The error judgment unit is used to determine whether the error between the current pose of the whole segment and the designed pose meets the preset error threshold. The correction control unit triggers a camera parameter adjustment strategy based on the error between the current pose and the designed pose of the entire segment, and completes the segment main splicing correction based on the camera parameter adjustment strategy.

Images