Shield segment bolt hole alignment deviation detection method based on feature matching
By using a feature-matching method and employing 3D laser scanning and algorithm processing, precise alignment detection of bolt holes in tunnel lining segments was achieved. This solved the problems of inaccurate detection and low efficiency in existing technologies, and enabled efficient and accurate assessment of bolt hole alignment deviations.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHANGHAI TUNNEL ENG CO LTD
- Filing Date
- 2026-02-24
- Publication Date
- 2026-05-29
AI Technical Summary
In shield tunnel construction, existing technologies rely on manual experience to detect the alignment of bolt holes in shield segments, which leads to inaccurate detection, low efficiency, lack of quantification, and safety risks.
A feature-matching-based method is adopted, which collects point cloud data through a 3D laser scanner, identifies bolt holes using the RANSAC plane fitting algorithm and Euclidean clustering algorithm, and performs bolt hole pairing and deviation calculation by combining the cylindrical fitting algorithm, thereby achieving accurate alignment detection of bolt holes.
It enables accurate quantitative detection of bolt hole alignment deviation, improves detection efficiency, avoids the impact of manual intervention on data accuracy, and makes detection records easy to store and retrieve.
Smart Images
Figure CN122107931A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of shield tunnel construction technology, and specifically relates to a method for detecting alignment deviation of shield tunnel segment bolt holes based on feature matching. Background Technology
[0002] In shield tunnel construction, the precise assembly of tunnel segments is the core of tunnel structural quality. The direct physical manifestation of proper assembly is the complete alignment of bolt holes between rings and segments, allowing bolts to pass through and tighten smoothly. The alignment accuracy of the bolt holes is the most critical and intuitive on-site indicator for assessing assembly quality and ensuring structural safety and waterproofing reliability.
[0003] Currently, the inspection of this crucial step relies almost entirely on manual experience: workers must use flashlights to illuminate the holes and judge the alignment between them by visual observation and experience. This method is extremely subjective, cannot be quantified, and has serious limitations: ① It can only qualitatively assess "whether it can be inserted," and cannot accurately quantify the alignment deviation; ② The inspection environment is dim and the viewing angle is limited, posing significant safety risks, and the results are difficult to record and trace. Therefore, we propose a feature-matching-based method for detecting alignment deviations of shield tunnel segment bolt holes to solve the above problems. Summary of the Invention
[0004] To address the aforementioned issues, this invention provides a method for detecting alignment deviations of bolt holes in tunnel lining segments based on feature matching. This method solves the problems of inaccuracy, low efficiency, and inability to quantify the assessment caused by relying on manual visual inspection to determine the alignment accuracy of bolt holes during tunnel lining segment assembly.
[0005] This invention is achieved through the following scheme: a method for detecting alignment deviation of bolt holes in tunnel lining segments based on feature matching, comprising the following steps: S1. Collect point cloud data of the plane containing the bolt holes between the segments to be assembled and the installed segments; S2. The collected point cloud data is processed by an algorithm and stitched together into a complete 3D point cloud dataset in a unified coordinate system. S3. Use the RANSAC plane fitting algorithm to identify and extract the planar point cloud data subsets where each segment bolt hole is located from the 3D point cloud dataset; S4. For all subsets of planar point cloud data, calculate their normal vector field, and combine the point cloud density changes and local curvature features on the plane to identify the circular hole regions on the plane. S5. Use the Euclidean clustering algorithm to segment the point cloud data of each circular hole region into independent point cloud clusters, and each point cloud cluster corresponds to a bolt hole; S6. Use a cylindrical fitting algorithm to fit each point cloud cluster into a cylindrical model, and output the direction vector of the cylindrical central axis of each bolt hole and the three-dimensional coordinates of the center point of the bottom surface of the cylinder. The direction vector of the cylindrical central axis is the orientation of the bolt hole, and the three-dimensional coordinates of the center point of the bottom surface of the cylinder is the position of the bolt hole. S7. Using the relative position and orientation relationship between the two bolt holes on the segment design model as the constraint basis, find the bolt holes on the installed segment plane that are closest to the bolt holes on the segment to be assembled and whose axial direction is approximately parallel to each other for pairing. S8. Calculate the alignment deviation of each pair of bolt holes, compare the alignment deviations to determine whether each pair of bolt holes is aligned correctly.
[0006] A further improvement of the feature matching-based method for detecting alignment deviation of bolt holes in tunnel segments is that the alignment deviation includes the Euclidean distance between the three-dimensional coordinates of the center points of the bottom surfaces of the cylinders of each pair of bolt holes and the spatial angle between the direction vectors of the central axes of the cylinders of each pair of bolt holes. The comparison methods for comparing alignment deviations include A and B: The comparison method A is: whether the Euclidean distance between the three-dimensional coordinates of the center points of the bottom surface of each pair of bolt holes is less than the set distance threshold; The comparison method B is: whether the spatial angle between the direction vectors of the central axis of each pair of bolt holes is less than a set angle threshold. The method for determining whether each pair of bolt holes is aligned is as follows: if all the results of the comparison methods are yes, then the alignment is deemed to be qualified; if the result of any comparison method is no, then the alignment is deemed to be unqualified.
[0007] A further improvement of the feature-matching-based method for detecting alignment deviation of bolt holes in tunnel segments is that the alignment deviation also includes the positional deviation of the two center points in each pair of bolt holes in the axial, radial and circumferential directions of the tunnel. The comparison method further includes: whether the positional deviations of the two center points in each pair of bolt holes in the axial, radial and circumferential directions along the tunnel are all less than the set positional deviation thresholds.
[0008] A further improvement of the method for detecting alignment deviation of bolt holes in tunnel lining segments based on feature matching in this invention is that, in step S1, a three-dimensional laser scanner is used to collect point cloud data through multi-station scanning.
[0009] A further improvement of the method for detecting alignment deviation of bolt holes in tunnel lining segments based on feature matching in this invention is that the installed segments include segments in the previous ring of segments that correspond to the segment to be assembled, as well as segments in the same ring of segments that are adjacent to the segment to be assembled.
[0010] A further improvement of the method for detecting alignment deviation of bolt holes in tunnel lining segments based on feature matching in this invention is that, before performing step S3, it also includes the steps of denoising and filtering outliers on the three-dimensional point cloud dataset.
[0011] A further improvement of the method for detecting alignment deviation of bolt holes in tunnel lining segments based on feature matching in this invention is that, in step S6, the direction vector of the cylindrical center axis of all bolt holes on the installed segments is uniformly adjusted to be perpendicular to the plane where the bolt holes are located and pointing to the interior of the corresponding segments; and the direction vector of the cylindrical center axis of all bolt holes on the segments to be assembled is uniformly adjusted to be perpendicular to the plane where the bolt holes are located and pointing to the plane where the corresponding bolt holes are located on the installed segments.
[0012] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention generates independent point cloud clusters by collecting, processing, extracting, identifying, and segmenting point cloud data of the plane containing bolt holes. It uses a cylindrical fitting algorithm to obtain the direction vector of the central axis of the cylinder and the three-dimensional coordinates of the center point of the bottom surface of the cylinder for each point cloud cluster. Then, it performs pairing, calculation, and comparison of bolt holes to detect whether each pair of bolt holes is aligned correctly. This intelligent detection method is highly efficient, avoids the impact of manual intervention on data accuracy, and the detection records are easy to store and retrieve. Attached Figure Description
[0013] Figure 1 A schematic flowchart of the alignment deviation detection method of the present invention is shown.
[0014] Figure 2 A schematic diagram of the assembly position of the segments to be assembled according to the present invention is shown.
[0015] In the diagram: 1. Segment to be assembled; 2. Segment in the same ring; 3. Segment in the previous ring; 4. Bolt hole; 5. Cylindrical model; 6. Center point of the bottom surface of the cylinder; 7. Direction vector of the central axis of the cylinder. Detailed Implementation
[0016] To address the problems of inaccuracy, low efficiency, and lack of quantifiable evaluation caused by relying on manual visual inspection to determine bolt hole alignment accuracy during shield tunnel segment assembly, this invention provides a feature-matching-based method for detecting bolt hole alignment deviations in shield tunnel segments. The following detailed description, in conjunction with accompanying drawings, illustrates this feature-matching-based method for detecting bolt hole alignment deviations in shield tunnel segments.
[0017] See Figures 1-2 As shown, a method for detecting alignment deviation of bolt holes in tunnel lining segments based on feature matching includes the following steps: S1. Collect point cloud data of the plane containing the bolt holes 4 between the segment to be assembled 1 and the installed segment; S2. The collected point cloud data is processed by an algorithm and stitched together into a complete 3D point cloud dataset in a unified coordinate system. S3. Use the RANSAC plane fitting algorithm to identify and extract the plane point cloud data subsets where each segment bolt hole 4 is located from the three-dimensional point cloud dataset; S4. For all subsets of planar point cloud data, calculate their normal vector field, and combine the point cloud density changes and local curvature features on the plane to identify the circular hole regions on the plane. S5. Use the Euclidean clustering algorithm to segment the point cloud data of each circular hole region into independent point cloud clusters, with each point cloud cluster corresponding to a bolt hole 4. S6. Use a cylindrical fitting algorithm to fit each point cloud cluster into a cylindrical model 5, and output the direction vector 7 of the cylindrical central axis of each bolt hole 4 and the three-dimensional coordinates of the center point 6 of the bottom surface of the cylinder. The direction vector 7 of the cylindrical central axis is the orientation of the bolt hole 4, and the three-dimensional coordinates of the center point 6 of the bottom surface of the cylinder is the position of the bolt hole 4. S7. Using the relative position and orientation relationship between the two bolt holes 4 on the segment design model as the constraint basis, find the bolt hole 4 that is closest to the bolt hole 4 on the segment 1 to be assembled and whose axial direction is approximately parallel on the plane corresponding to the installed segment and the segment 1 to be assembled, and match them. S8. Calculate the alignment deviation of each pair of bolt holes 4, and compare the alignment deviations to determine whether each pair of bolt holes 4 is aligned correctly.
[0018] By collecting, processing, extracting, identifying, and segmenting the point cloud data of the plane containing bolt holes 4, independent point cloud clusters are generated. A cylindrical fitting algorithm is used to obtain the direction vector 7 of the central axis of the cylinder and the three-dimensional coordinates of the center point 6 of the bottom surface of the cylinder for each point cloud cluster. Then, the bolt holes 4 are paired, calculated, and compared to detect whether each pair of bolt holes 4 is aligned. This intelligent detection method is highly efficient, avoids the impact of manual intervention on the accuracy of the data, and the detection records are easy to store and retrieve.
[0019] The alignment deviation includes the Euclidean distance between the three-dimensional coordinates of the center point 6 on the bottom surface of each pair of bolt holes 4 cylinders and the spatial angle between the direction vectors of the central axis of each pair of bolt holes 4 cylinders. The comparison methods for reconciling alignment deviations include A and B: Comparison method A is: whether the Euclidean distance between the three-dimensional coordinates of the center point 6 of the bottom surface of each pair of bolt holes 4 is less than the set distance threshold (2mm is recommended); Comparison method B is: whether the spatial angle between the directional vectors of the four cylindrical center axes of each pair of bolt holes is less than the set angle threshold (recommended to be 0.1°); The method for determining whether each pair of bolt holes 4 is aligned is as follows: if all comparison methods result in "yes", then the alignment is considered acceptable; if any comparison method results in "no", then the alignment is considered unacceptable.
[0020] By adopting the above design, the alignment deviation can be accurately quantified, and the positional deviation can directly reflect the misalignment between the segments, while the angle deviation can directly reflect the degree of torsion or misalignment of the bolt hole 4.
[0021] The alignment deviation also includes the positional deviation of the two center points in each pair of bolt holes 4 in the axial, radial and circumferential directions along the tunnel; The comparison method also includes: whether the positional deviations of the two center points in each pair of bolt holes 4 in the axial, radial and circumferential directions along the tunnel are all less than the set positional deviation threshold (1mm is recommended).
[0022] By adopting the above design, the alignment deviation can be accurately quantified, which greatly improves the accuracy of alignment of each pair of bolt holes.
[0023] In step S1, point cloud data is collected using a 3D laser scanner through multi-station scanning.
[0024] By adopting the above design, efficient and accurate 3D data acquisition can be achieved, greatly improving the accuracy of the acquired data.
[0025] The installed segments include the segments in the previous ring of segments 3 that correspond to the segments to be assembled 1, as well as the segments in the same ring of segments 2 that are adjacent to the segments to be assembled 1.
[0026] Before performing step S3, the process also includes the steps of denoising the 3D point cloud dataset and filtering outliers.
[0027] By adopting the above design and using voxelized mesh for downsampling, the amount of data is reduced while ensuring feature accuracy, thereby improving the speed of subsequent processing.
[0028] Specifically, during step S6, the direction vector of the cylindrical center axis of all bolt holes 4 on the installed pipe segment is uniformly adjusted to be perpendicular to the plane where the bolt hole 4 is located and points to the inside of the corresponding pipe segment; the direction vector of the cylindrical center axis of all bolt holes 4 on the pipe segment 1 to be assembled is uniformly adjusted to be perpendicular to the plane where the bolt hole 4 is located and points to the plane where the corresponding bolt hole 4 is located on the installed pipe segment.
[0029] By adopting the above design, the direction vector of the cylindrical center axis of all bolt holes is unified, which ensures that the sign of a certain component of all axis vectors is the same, facilitating subsequent calculation data.
[0030] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.
[0031] The present invention has been described in detail above with reference to the accompanying drawings and embodiments. Those skilled in the art can make various modifications to the present invention based on the above description. Therefore, certain details in the embodiments should not be construed as limiting the present invention, and the scope of protection of the present invention shall be defined by the appended claims.
Claims
1. A method for detecting alignment deviation of bolt holes in tunnel lining segments based on feature matching, characterized in that, Includes the following steps: S1. Collect point cloud data of the plane containing the bolt holes between the segments to be assembled and the installed segments; S2. The collected point cloud data is processed by an algorithm and stitched together into a complete 3D point cloud dataset in a unified coordinate system. S3. Use the RANSAC plane fitting algorithm to identify and extract the planar point cloud data subsets where each segment bolt hole is located from the 3D point cloud dataset; S4. For all subsets of planar point cloud data, calculate their normal vector field, and combine the point cloud density changes and local curvature features on the plane to identify the circular hole regions on the plane. S5. Use the Euclidean clustering algorithm to segment the point cloud data of each circular hole region into independent point cloud clusters, and each point cloud cluster corresponds to a bolt hole; S6. Use a cylindrical fitting algorithm to fit each point cloud cluster into a cylindrical model, and output the direction vector of the cylindrical central axis of each bolt hole and the three-dimensional coordinates of the center point of the bottom surface of the cylinder. The direction vector of the cylindrical central axis is the orientation of the bolt hole, and the three-dimensional coordinates of the center point of the bottom surface of the cylinder is the position of the bolt hole. S7. Using the relative position and orientation relationship between the two bolt holes on the segment design model as the constraint basis, find the bolt holes on the installed segment plane that are closest to the bolt holes on the segment to be assembled and whose axial direction is approximately parallel to each other for pairing. S8. Calculate the alignment deviation of each pair of bolt holes, compare the alignment deviations to determine whether each pair of bolt holes is aligned correctly.
2. The method for detecting alignment deviation of shield tunnel segment bolt holes based on feature matching as described in claim 1, characterized in that, The alignment deviation includes the Euclidean distance between the three-dimensional coordinates of the center points of the bottom surfaces of each pair of bolt hole cylinders and the spatial angle between the direction vectors of the central axis of each pair of bolt hole cylinders. The comparison methods for comparing alignment deviations include A and B: The comparison method A is: whether the Euclidean distance between the three-dimensional coordinates of the center points of the bottom surface of each pair of bolt holes is less than the set distance threshold; The comparison method B is: whether the spatial angle between the direction vectors of the central axis of each pair of bolt holes is less than a set angle threshold. The method for determining whether each pair of bolt holes is aligned is as follows: if all the results of the comparison methods are yes, then the alignment is deemed to be qualified; if the result of any comparison method is no, then the alignment is deemed to be unqualified.
3. The method for detecting alignment deviation of shield tunnel segment bolt holes based on feature matching as described in claim 2, characterized in that, The alignment deviation also includes the positional deviation of the two center points in each pair of bolt holes in the axial, radial and circumferential directions along the tunnel; The comparison method further includes: whether the positional deviations of the two center points in each pair of bolt holes in the axial, radial and circumferential directions along the tunnel are all less than the set positional deviation thresholds.
4. The method for detecting alignment deviation of shield tunnel segment bolt holes based on feature matching as described in claim 1, characterized in that, In step S1, point cloud data is collected using a 3D laser scanner through multi-station scanning.
5. The method for detecting alignment deviation of shield tunnel segment bolt holes based on feature matching as described in claim 1, characterized in that, The installed segments include the segments in the previous ring that correspond to the segments to be assembled, as well as the segments in the same ring that are adjacent to the segments to be assembled.
6. The method for detecting alignment deviation of shield tunnel segment bolt holes based on feature matching as described in claim 1, characterized in that, Before performing step S3, the steps include: denoising the 3D point cloud dataset and filtering outliers.
7. The method for detecting alignment deviation of shield tunnel segment bolt holes based on feature matching as described in claim 1, characterized in that, When performing step S6, the direction vector of the cylindrical center axis of all bolt holes on the installed segments is uniformly adjusted to be perpendicular to the plane where the bolt holes are located and pointing to the inside of the corresponding segments. The direction vector of the cylindrical center axis of all bolt holes on the segments to be assembled is uniformly adjusted to be perpendicular to the plane where the bolt holes are located and pointing to the plane where the corresponding bolt holes are located on the installed segments.