Arch bridge cantilever attitude and line shape real-time monitoring method based on unmanned aerial vehicle ground-air conversion

Abstract

The invention discloses an unmanned aerial vehicle ground-to-air conversion-based arch bridge cantilever attitude and line shape real-time monitoring method, which comprises the following steps that: after segments are assembled on the ground, a plurality of prisms which are not positioned on the same straight line and a plurality of targets which are not positioned on the same straight line are arranged on the upper surfaces of the segments; three-dimensional point clouds of the segments are constructed, and coordinates of the three-dimensional point clouds are corrected through the targets; constructing a key point set, constructing a prism coordinate system (XYZ) L0, and converting coordinates of the key point set to the prism coordinate system (XYZ) L0; when the segments are spliced in the air, whether the prisms are subjected to construction disturbance deviation or not is checked, and the spatial positions and postures of chord members and flange holes of the segments are obtained according to the prism disturbance deviation condition. According to the invention, the world coordinates of the chord member pipe orifice center and the flange hole center of the main arch segment can be accurately obtained, and the spatial position and posture of the segment can be effectively reflected.

Description

Technical Field

[0001] The present invention relates to the field of bridge construction, and in particular to a method for real-time monitoring of the posture and linear shape of an arch bridge cantilever based on ground-to-air conversion of an unmanned aerial vehicle. Background Art

[0002] High-precision measurement and control of the linear shape of large-span steel tube concrete arch bridges are the key to the safety of the construction process and the quality of the arch. Usually, the tension of the inclined buckle cable is required to adjust the suspended posture of the arch rib segment and the linear shape of the cantilever to achieve the preset elevation and mileage and avoid lateral deviation. If the segment is displaced during splicing, it will be difficult to correct the error during the construction and installation of subsequent segments, which will eventually affect the smooth closure of the main arch and cause permanent defects in the stress-free state of the arch axis. In addition, if the spatial posture of the segment is inaccurate, there will inevitably be gaps in the flange splicing when the segments are spliced ​​together, and even the flange holes often do not match, requiring repeated hole expansion, and high-altitude operations are difficult.

[0003] To identify the splicing posture and cantilever line shape of the arch bridge segment, the spatial position of the pipe ends of each chord and the splicing flange hole position of the chord must be determined at the same time. However, the current monitoring accuracy of the splicing posture of the main arch segment of the large-span steel tube concrete arch bridge is low. The segment head position is controlled by the waistline measurement method, and only the spatial position of the chord where the measuring point is located can be monitored. Usually, a segment contains four axial main chords. Only the outer surface of one of the chords (steel tube) is measured, which cannot reflect the position of the pipe ends of the other chords, especially the spatial position (xyz coordinates) and posture (3 spatial inclination angles) of the splicing flange. Therefore, it cannot reflect the spatial position and posture of the entire segment.

[0004] Therefore, in order to solve the above problems, a real-time monitoring method for the cantilever posture and linearity of an arch bridge based on UAV ground-to-air conversion is needed, which can accurately obtain the world coordinates of the chord pipe mouth center and the flange hole center of the main arch segment, and effectively reflect the spatial position and posture of the segment. Summary of the invention

[0005] In view of this, the purpose of the present invention is to overcome the defects in the prior art and provide a real-time monitoring method for the cantilever posture and line shape of an arch bridge based on ground-to-air conversion of an unmanned aerial vehicle, which can accurately obtain the world coordinates of the chord pipe mouth center and the flange hole center of the main arch segment, and effectively reflect the spatial position and posture of the segment.

[0006] The method for real-time monitoring of the posture and linear shape of an arch bridge cantilever based on ground-to-air conversion of an unmanned aerial vehicle of the present invention comprises:

[0007] After the segments are assembled on the ground, a number of prisms and a number of targets that are not in the same straight line are arranged on the upper surface of the segments;

[0008] Construct the three-dimensional point cloud of the segment and use the target to correct the coordinates of the three-dimensional point cloud;

[0009] Construct key point set and build prism coordinate system (XYZ) L0 , transform the coordinates of the key point set to the prism coordinate system (XYZ) L0 ;

[0010] When the segments are being spliced ​​in the air, check whether the prism is disturbed by the construction and deviated. If not, build the prism coordinate system (XYZ) of the current segment. L1 , calculate its coordinate system with the prism (XYZ) L0 The rotation matrix is ​​used to transform all coordinates in the key point set into the world coordinate system to obtain the spatial position and posture of the chord and flange hole of the segment; if yes, the spatial relationship between the prism system and the target system is reconstructed to obtain the spatial position and posture of the chord and flange hole of the segment.

[0011] Furthermore, the three-dimensional point cloud coordinates are corrected using the target, specifically including:

[0012] Based on the four non-coplanar feature points in space, a spatial tetrahedron unit is established, and the six side lengths of the tetrahedron unit are measured once in the three-dimensional point cloud, which is recorded as Lm. 1～6 , and then measure it once on the corresponding actual object, recorded as L 1～6 , then use L 1～6 / Lm 1～6 , get the conversion coefficients k1~k6, take the average of the conversion coefficients k1~k6, get the average conversion coefficient km, scale the coordinates of all 3D point clouds according to km, then measure the scaled point clouds, and get the coordinates of any point in the point cloud at the real scale:

[0013] (x,y,z)=km(x m ,y m ,z m );

[0014] Among them, (x m ,y m ,z m ) and (x, y, z) are the 3D point cloud coordinates and the real-scale coordinates of a point in the 3D point cloud respectively.

[0015] Furthermore, the key point set includes the coordinates of the prism center point, the target center point, the chord pipe mouth center point and the flange hole center point in the three-dimensional point cloud.

[0016] Further, construct the prism coordinate system (XYZ) L0 , specifically including:

[0017] Measure the world coordinates of the center points of the three prisms, recorded as P1 (x1, y1, z1), P2 (x2, y2, z2), and P3 (x3, y3, z3); ​​calculate the distance between the three points, and the distance between P1 and P2 is recorded as L 12 , the distance between P2 and P3 is denoted as L 23 , the distance between P1 and P3 is denoted as L 13 ;

[0018] In the point cloud model, the point cloud of the outer edge of the prism is screened out, and the noise points on the non-circular surface are filtered out using the RANSAC method. Then, the three-dimensional coordinates of the center of the prism in the point cloud model are obtained by spatial circle center fitting. The specific process is as follows:

[0019] Randomly select 3 coordinate points from all outer edge point clouds, calculate the circle equation obtained by the 3 points, and obtain the center coordinates p and radius r;

[0020] Set a distance threshold t, traverse the distance between the coordinate points in all outer edge point clouds and all circle center coordinates p, and use the difference between the distance and the radius r as the deviation value. Mark the coordinate points with deviation values ​​less than the distance threshold t as inliers, and those exceeding the distance threshold t as outliers.

[0021] Repeat the above process continuously, record the number of interior points and interior points obtained each time, and repeat N times to end the iteration;

[0022] Select the iteration result with the largest number of interior points during the iteration process, remove the outliers in the result, and make the final space circle fitting based on the interior points of the result;

[0023] Randomly select two points on the outer edge and compare them with the center point of the fitted prism edge (X O ,Y O ,Z O ) are connected into two intersecting vectors, the plane normal vector is obtained by cross-producting the two vectors, the normal vectors obtained by different random vectors are iteratively calculated, the vector component deviation threshold is set, the normal vectors exceeding the deviation threshold are eliminated, and the mean of the remaining normal vectors is used as the normal vector F of the final prism outer contour circle;

[0024] According to the known depth d of the prism center point, (X O ,Y O ,Z O ) is translated inwards by a distance d in the direction of the normal vector F, and the final prism optical measurement center point coordinates (X OL ,Y OL ,Z OL ); The center point coordinates (X OL ,Y OL ,Z OL ) 1～3 , establish the prism coordinate system (XYZ)L0 .

[0025] Furthermore, the least squares method is used to fit the center of the space circle, and the expansion term of the center of the space circle is calculated once:

[0026]

[0027] Among them, X i , Y i , Z i is the spatial coordinate of the edge point of the outer contour of the i-th prism in the point cloud model, X O , Y O , Z O The coordinates of the circle center obtained by a single sampling; R o Indicates the radius of the outer contour circle of the prism.

[0028] Furthermore, check whether the prism is disturbed or misaligned during construction, including:

[0029] When placing the segments on the ground, the coordinates of the center points of the three prisms are measured for the first time, and the side lengths of the triangles formed are calculated and recorded; then, during the aerial assembly stage, the coordinates of the center points of the three prisms are measured again, and the absolute value of the difference between the new triangle side length and the side length calculated on the ground is calculated. If the absolute value exceeds the threshold τ, the prism is displaced by construction disturbance; otherwise, the prism is not displaced by construction disturbance.

[0030] Furthermore, the spatial relationship between the prism system and the target system is reconstructed, including:

[0031] When the segments are assembled on the ground, the target coordinate system (XYZ) is established with the target center point. B0 , and convert all key point set coordinates to (XYZ) B0 In the example, the coordinate set {P} B0 ;

[0032] The three prisms and three targets on the spliced ​​arch rib segments were re-scanned and photographed, and dense 3D reconstruction was performed on them. The outer contour point clouds were fitted to obtain the coordinates of their central point clouds, and then corrected according to the actual spacing of the targets to obtain the point cloud coordinates of the center points of the prisms and targets in the new 3D model.

[0033] Establish the prism point cloud coordinate system (XYZ) with the point cloud coordinates of the prism center point of the new 3D model LD , by measuring the real spatial coordinates of the prism, establish the prism coordinate system (XYZ) L1 , assuming the key point set is (XYZ) LD and (XYZ) L1 The coordinates in both coordinate systems are the same;

[0034] The coordinate set {P}B0 Convert to Prism point cloud coordinate system (XYZ) LD , and then from the prism point cloud coordinate system (XYZ) LD Convert to prism coordinate system (XYZ) L1 , and finally transformed to the global coordinate system.

[0035] Furthermore, the method further includes: adjusting the segment suspension posture:

[0036] Based on the world coordinates of the key point set, it is compared with the design coordinates to obtain the deviation of the segment suspension posture, and the current segment splicing posture is corrected by adjusting the cable tension.

[0037] The beneficial effects of the present invention are as follows: the present invention discloses a method for real-time monitoring of the cantilever posture and line shape of an arch bridge based on ground-to-air conversion of an unmanned aerial vehicle, which only requires the use of an automatic measuring robot to scan and measure the three prisms of the segment, and the measurement is rapid; the relative relationship (local coordinates) between the key points and the prisms recorded on the ground in advance can be used to quickly feedback the world coordinates of all key points on the segment. The prisms left on the outer surface of the steel pipe can accurately feedback the spatial posture and cantilever line shape of the entire main arch segment, and the operation is simple; for the special case where the prism is disturbed, it is only necessary to send out a drone to reshoot in the cantilever assembly stage, and quickly reconstruct the solution, and the process is simple. It has strong adaptability to complex construction conditions and can correct back to accurate recognition results after the prism is disturbed and shifted. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] The present invention will be further described below in conjunction with the accompanying drawings and embodiments:

[0039] Figure 1 It is a schematic diagram of the arrangement of the prism and the target of the present invention;

[0040] Figure 2 It is a schematic diagram of the drone photography of the present invention;

[0041] Among them, 1-main arch segment, 2-prism, 3-target, 4-drone lens. DETAILED DESCRIPTION

[0042] The present invention is further described below in conjunction with the accompanying drawings, as shown in the drawings:

[0043] This embodiment discloses a method for real-time monitoring of the posture and linear shape of an arch bridge cantilever based on ground-to-air conversion of an unmanned aerial vehicle, comprising the following steps:

[0044] After the segments are assembled on the ground, a number of prisms and a number of targets that are not in the same straight line are arranged on the upper surface of the segments;

[0045] Construct the three-dimensional point cloud of the segment and use the target to correct the coordinates of the three-dimensional point cloud;

[0046] Construct key point set and build prism coordinate system (XYZ) L0 , transform the coordinates of the key point set to the prism coordinate system (XYZ) L0 ;

[0047] When the segments are being spliced ​​in the air, check whether the prism is disturbed by the construction and deviated. If not, build the prism coordinate system (XYZ) of the current segment. L1 , calculate its coordinate system with the prism (XYZ) L0 The rotation matrix is ​​used to transform all coordinates in the key point set into the world coordinate system to obtain the spatial position and posture of the chord and flange hole of the segment; if yes, the spatial relationship between the prism system and the target system is reconstructed to obtain the spatial position and posture of the chord and flange hole of the segment.

[0048] In this embodiment, Figure 1 As shown, after the segment is assembled on the ground, three prisms that are not in the same straight line are fixed on the upper surface of the segment, and the spatial xyz coordinates of the center points of the prisms are measured using a total station or an automatic measuring robot, and the actual distance between the two prisms is calculated; in addition, three targets are horizontally pasted near each prism; similarly, the three targets are not in the same straight line.

[0049] Through drone photography, multi-angle segment port images are taken at both ends of the segment placed on the ground; and three targets set on the segment surface are photographed at multiple angles directly above the segment; the distance between the targets is measured and recorded; the SFM or SLAM 3D reconstruction method is used to convert the drone image into a 3D visual point cloud of the segment; the center points of the three targets are picked up, and the distance between the three center points in the 3D model is calculated. The ratio of the real distance to the pixel distance is used as the coefficient to correct all point cloud coordinates:

[0050] Based on the four non-coplanar feature points in space, a spatial tetrahedron unit is established, and the six side lengths of the tetrahedron unit are measured once in the three-dimensional point cloud, which is recorded as Lm. 1～6 , and then measure it once on the corresponding actual object, recorded as L 1～6 , then use L 1～6 / Lm 1～6 , get the conversion coefficients k1~k6, take the average of the conversion coefficients k1~k6, get the average conversion coefficient km, scale the coordinates of all point clouds according to km, and then measure the scaled point clouds to get the coordinates of any point in the point cloud model at the real scale; where, (x,y,z)=k m (x m ,y m ,z m );(xm ,y m ,z m ) and (x, y, z) are the three-dimensional reconstructed coordinates and the real-scale coordinates of a point in the model respectively.

[0051] In this embodiment, the edge feature points of the prism, target, and flange hole are screened out in the visual point cloud, and the coordinates of the center point of the prism, the center point of the target, the center point of the chord pipe mouth, and the center point of all flange holes in the three-dimensional point cloud model are estimated by the least squares method, which is recorded as a "key point set";

[0052] According to the corrected pixel space coordinates of the three prism center points, the original prism local coordinate system (XYZ) is established L0 , and use the existing visual geometry technology to convert the coordinates of all corrected key point sets to the original prism coordinate system (XYZ) L0 Next. Construct the prism coordinate system (XYZ) L0 , specifically including:

[0053] Measure the world coordinates of the center points of the three prisms, recorded as P1 (x1, y1, z1), P2 (x2, y2, z2), and P3 (x3, y3, z3); ​​calculate the distance between the three points, and the distance between P1 and P2 is recorded as L 12 , the distance between P2 and P3 is denoted as L 23 , the distance between P1 and P3 is denoted as L 13 ;

[0054] In the visual point cloud model generated by the image, the point cloud of the outer edge of the prism is screened out. Considering that the effect of the least squares fitting method is affected by noise data, image occlusion, etc., the RANSAC method is used to filter out the noise points on the non-circular surface, and then the three-dimensional coordinates of the center of the prism in the point cloud model are obtained by spatial circle center fitting. The specific process is as follows:

[0055] Randomly select 3 coordinate points from all outer edge point clouds, calculate the circle equation obtained by the 3 points, and obtain the center coordinates p and radius r;

[0056] Set a distance threshold t, traverse the distance between the coordinate points in all outer edge point clouds and all circle center coordinates p, and use the difference between the distance and the radius r as the deviation value. Mark the coordinate points with deviation values ​​less than the distance threshold t as inliers, and those exceeding the distance threshold t as outliers.

[0057] Repeat the above process continuously, record the number of interior points and interior points obtained each time, and repeat N times to end the iteration;

[0058] Select the iteration result with the largest number of interior points in the iteration process, remove the outliers in the result, and make the final space circle fitting based on the interior points of the result; use the least squares method to fit the center of the space circle, and calculate the expansion term of the center of the space circle once:

[0059]

[0060] Among them, X i , Y i , Z i is the spatial coordinate of the edge point of the outer contour of the i-th prism in the point cloud model, X O , Y O , Z O The coordinates of the circle center obtained by a single sampling; R o Indicates the radius of the outer contour circle of the prism. The above process is also applicable to the fitting of the center of the chord pipe mouth and the center of the flange bolt hole.

[0061] Randomly select two points on the outer edge and compare them with the center point of the fitted prism edge (X O ,Y O ,Z O ) are connected into two intersecting vectors, the plane normal vector is obtained by cross-producting the two vectors, the normal vector obtained by different random vectors is iteratively calculated, the vector component deviation threshold is set, the normal vector exceeding the deviation threshold is eliminated, and the mean of the remaining normal vectors is used as the normal vector F of the final prism outer contour circle; wherein the distance threshold and the deviation threshold can be set according to the actual working conditions;

[0062] According to the known depth d of the prism center point, (X O ,Y O ,Z O ) is translated inwards by a distance d in the direction of the normal vector F, and the final prism optical measurement center point coordinates (X OL ,Y OL ,Z OL ); The center point coordinates (X OL ,Y OL ,Z OL ) 1～3 , establish the prism coordinate system (XYZ) L0 .

[0063] In this embodiment, after the segment is hoisted to the installation position and initially fixed, the global coordinates of the center points of the three prisms are measured again using the SFM 3D reconstruction method, and the prism spacing in the current state is calculated; the prism spacing is compared with the prism spacing measured on the ground to check whether the prism is disturbed and deviated by the construction, specifically including:

[0064] When placing the segments on the ground, the coordinates of the center points of the three prisms are measured for the first time, and the side lengths of the triangles formed are calculated and recorded. Then, during the aerial assembly stage, the coordinates of the center points of the three prisms are measured again, and the absolute value of the difference between the new triangle side length and the side length calculated on the ground is calculated. If the absolute value exceeds the threshold of 1 cm, it means that the prism has been affected by collisions, wire rope strangulation, etc. during the construction process, that is, the prism has been disturbed and displaced by the construction. Otherwise, the prism has not been disturbed and displaced by the construction.

[0065] In this embodiment, if the three prisms are not disturbed by the construction, the prism coordinate system (XYZ) of the current segment is directly established. L1 , calculate its coordinate system with the original prism (XYZ) L0 The rotation matrix T is used to transform all coordinates in the key point set into the world coordinate system, and the world coordinates of the chord nozzle center and the flange hole center of the main arch segment can be obtained, thereby obtaining the spatial position and posture of the chord and flange hole of the segment.

[0066] If any prism is deformed by the wire rope, the spatial relationship between the prism system and the target system is reconstructed:

[0067] Since the local coordinate system (XYZ) of the prism L0 If the target has failed, the target coordinate system (XYZ) is established with the target center point when the steel truss segment is assembled on the ground. B0 , and convert all key point set coordinates to (XYZ) B0 In the example, the coordinate set {P} B0 , which is used to restore the original relative position relationship of the key point set. Because the target is firmly attached and not easily disturbed, it can be considered that the {P} based on the target B0 It remains unchanged throughout the construction process;

[0068] like Figure 2 As shown, a drone is dispatched to rescan and photograph the three prisms and three targets on the spliced ​​arch rib segment, and dense 3D reconstruction is performed on them, and the outer contour point cloud is fitted to obtain the coordinates of the center point cloud, and then the point cloud coordinates of the center points of the prisms and targets in the new 3D model are re-corrected according to the actual spacing of the targets; wherein, the re-correction principle is similar to the principle of correcting the 3D point cloud coordinates using the targets mentioned above, and will not be repeated here;

[0069] Establish the prism point cloud coordinate system (XYZ) with the point cloud coordinates of the prism center point of the new 3D model LD , by measuring the real spatial coordinates of the prism, establish the prism coordinate system (XYZ) L1 , assuming the key point set is (XYZ) LD and (XYZ) L1The coordinates in the two coordinate systems are the same; it should be noted that the coordinate system (XYZ) B0 、(XYZ) LD and (XYZ) L1 Construction principle and coordinate system (XYZ) L0 The construction principle is similar, so I won’t go into details here;

[0070] The coordinate set {P} B0 Convert to Prism point cloud coordinate system (XYZ) LD , and then from the prism point cloud coordinate system (XYZ) LD Convert to prism coordinate system (XYZ) L1 , and finally converted to the global coordinate system. Among them, the above-mentioned coordinate conversion can be implemented by using the existing conversion matrix method, which will not be repeated here. The global coordinate system is the world coordinate system, and the world coordinate system can be constructed by using the existing geographic information system (GIS) technology.

[0071] Through the above processing, after conversion to the global coordinate system, the actual coordinates of all points can be obtained under the new arch rib segment posture, thereby obtaining the spatial position and posture of the chord and flange hole of the segment.

[0072] In this embodiment, the present invention further comprises: adjusting the segment suspension posture:

[0073] Based on the world coordinates of the key point set, the design coordinates are compared with the design coordinates to obtain the deviation of the segment suspension posture, and the current segment splicing posture is corrected by adjusting the cable tension. The design coordinates are given by the design drawings.

[0074] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution of the present invention, which should be included in the scope of the claims of the present invention.

Claims

1. A method for real-time monitoring of the cantilever posture and linearity of an arch bridge based on ground-to-air conversion of an unmanned aerial vehicle, characterized in that: include: After the segments are assembled on the ground, a number of prisms and a number of targets that are not in the same straight line are arranged on the upper surface of the segments; Construct the three-dimensional point cloud of the segment and use the target to correct the coordinates of the three-dimensional point cloud; Construct key point set and build prism coordinate system (XYZ) L0 , transform the coordinates of the key point set to the prism coordinate system (XYZ) L0 ; When the segments are being spliced ​​in the air, check whether the prism is disturbed by the construction and deviated. If not, build the prism coordinate system (XYZ) of the current segment. L1 , calculate its coordinate system with the prism (XYZ) L0 The rotation matrix is ​​used to transform all coordinates in the key point set into the world coordinate system to obtain the spatial position and posture of the segment chord and flange hole; If so, the spatial relationship between the prism system and the target system is reconstructed to obtain the spatial position and posture of the chord and flange hole of the segment.

2. The method for real-time monitoring of the cantilever posture and linearity of an arch bridge based on ground-to-air conversion of an unmanned aerial vehicle according to claim 1 is characterized in that: Use the target to correct the 3D point cloud coordinates, including: Based on the four non-coplanar feature points in space, a spatial tetrahedron unit is established, and the six side lengths of the tetrahedron unit are measured once in the three-dimensional point cloud, which is recorded as Lm. 1～6 , and then measure it once on the corresponding actual object, recorded as L 1～6 , then use L 1～6 / Lm 1～6 , get the conversion coefficients k1~k6, take the average of the conversion coefficients k1~k6, get the average conversion coefficient km, scale the coordinates of all 3D point clouds according to km, then measure the scaled point clouds, and get the coordinates of any point in the point cloud at the real scale: (x,y,z)=km(x m ,y m ,from m ); Among them, (x m ,y m ,z m ) and (x, y, z) are the 3D point cloud coordinates and the real-scale coordinates of a point in the 3D point cloud respectively.

3. The method for real-time monitoring of the cantilever posture and linearity of an arch bridge based on ground-to-air conversion of an unmanned aerial vehicle according to claim 1 is characterized in that: The key point set includes the coordinates of the prism center point, the target center point, the chord pipe mouth center point and the flange hole center point in the three-dimensional point cloud.

4. The method for real-time monitoring of the cantilever posture and linearity of an arch bridge based on ground-to-air conversion of an unmanned aerial vehicle according to claim 1 is characterized in that: Constructing the prism coordinate system (XYZ) L0 , specifically including: Measure the world coordinates of the center points of the three prisms, recorded as P1 (x1, y1, z1), P2 (x2, y2, z2), and P3 (x3, y3, z3); ​​calculate the distance between the three points, and the distance between P1 and P2 is recorded as L 12 , the distance between P2 and P3 is denoted as L 23 , the distance between P1 and P3 is denoted as L 13 ; In the point cloud model, the point cloud of the outer edge of the prism is screened out, and the noise points on the non-circular surface are filtered out using the RANSAC method. Then, the three-dimensional coordinates of the center of the prism in the point cloud model are obtained by spatial circle center fitting. The specific process is as follows: Randomly select 3 coordinate points from all outer edge point clouds, calculate the circle equation obtained by the 3 points, and obtain the center coordinates p and radius r; Set a distance threshold t, traverse the distance between the coordinate points in all outer edge point clouds and all circle center coordinates p, and use the difference between the distance and the radius r as the deviation value. Mark the coordinate points with deviation values ​​less than the distance threshold t as inliers, and those exceeding the distance threshold t as outliers. Repeat the above process continuously, record the number of interior points and interior points obtained each time, and repeat N times to end the iteration; Select the iteration result with the largest number of interior points during the iteration process, remove the outliers in the result, and make the final space circle fitting based on the interior points of the result; Randomly select two points on the outer edge and compare them with the center point of the fitted prism edge (X O ,Y O ,Z O ) are connected into two intersecting vectors, the plane normal vector is obtained by cross-producting the two vectors, the normal vectors obtained by different random vectors are iteratively calculated, the vector component deviation threshold is set, the normal vectors exceeding the deviation threshold are eliminated, and the mean of the remaining normal vectors is used as the normal vector F of the final prism outer contour circle; According to the known depth d of the prism center point, (X O ,Y O ,Z O ) is translated inwards by a distance d in the direction of the normal vector F, and the final prism optical measurement center point coordinates (X OL ,Y OL ,Z OL ); The center point coordinates (X OL ,Y OL ,Z OL ) 1～3 , establish the prism coordinate system (XYZ) L0 .

5. The method for real-time monitoring of the cantilever posture and linearity of an arch bridge based on ground-to-air conversion of an unmanned aerial vehicle according to claim 4 is characterized in that: Use the least squares method to fit the center of the space circle and calculate the expansion term of the center of the space circle once: Among them, X i , Y i , Z i is the spatial coordinate of the edge point of the outer contour of the i-th prism in the point cloud model, X O , Y O , Z O The coordinates of the circle center obtained by a single sampling; R o Indicates the radius of the outer contour circle of the prism.

6. The method for real-time monitoring of the cantilever posture and linearity of an arch bridge based on ground-to-air conversion of an unmanned aerial vehicle according to claim 1 is characterized in that: Check whether the prism is displaced due to construction disturbance, including: When placing the segments on the ground, the coordinates of the center points of the three prisms are measured for the first time, and the side lengths of the triangles formed are calculated and recorded; then, during the aerial assembly stage, the coordinates of the center points of the three prisms are measured again, and the absolute value of the difference between the new triangle side length and the side length calculated on the ground is calculated. If the absolute value exceeds the threshold τ, the prism is displaced by construction disturbance; otherwise, the prism is not displaced by construction disturbance.

7. The method for real-time monitoring of the cantilever posture and linearity of an arch bridge based on ground-to-air conversion of an unmanned aerial vehicle according to claim 1 is characterized in that: Reconstruct the spatial relationship between the prism system and the target system, including: When the segments are assembled on the ground, the target coordinate system (XYZ) is established with the target center point. B0 , and convert all key point set coordinates to (XYZ) B0 In the example, the coordinate set {P} B0 ; The three prisms and three targets on the spliced ​​arch rib segments were re-scanned and photographed, and dense 3D reconstruction was performed on them. The outer contour point clouds were fitted to obtain the coordinates of their central point clouds, and then corrected according to the actual spacing of the targets to obtain the point cloud coordinates of the center points of the prisms and targets in the new 3D model. Establish the prism point cloud coordinate system (XYZ) with the point cloud coordinates of the prism center point of the new 3D model LD , by measuring the real spatial coordinates of the prism, establish the prism coordinate system (XYZ) L1 , assuming the key point set is (XYZ) LD and (XYZ) L1 The coordinates in both coordinate systems are the same; The coordinate set {P} B0 Convert to Prism point cloud coordinate system (XYZ) LD , and then from the prism point cloud coordinate system (XYZ) LD Convert to prism coordinate system (XYZ) L1 , and finally transformed to the global coordinate system.

8. The method for real-time monitoring of the cantilever posture and linearity of an arch bridge based on ground-to-air conversion of an unmanned aerial vehicle according to claim 1 is characterized by: Also includes: Adjust the segment suspension posture: Based on the world coordinates of the key point set, it is compared with the design coordinates to obtain the deviation of the segment suspension posture, and the current segment splicing posture is corrected by adjusting the cable tension.

Images