A method for evaluating the quality of pipe segment assembly based on three-dimensional point cloud

Abstract

The application discloses a pipe piece assembling quality evaluation method based on three-dimensional point clouds, and relates to the technical field of visual measurement, comprising the following steps: comparing the assembling structure consistency of a reference structure candidate set and assembling point cloud data, obtaining an optimal assembling scheme, correlating and mapping the boundary positions of each pipe piece in the optimal assembling scheme with the corresponding boundary positions in the assembling point cloud data to generate pipe piece position offset relations; based on the pipe piece position offset relations, simulating the splicing of the optimal assembling scheme to generate an assembling path accessibility table, evaluating the assembling quality of the assembling path accessibility table, dividing quality grades, and generating an assembling quality evaluation table. The application improves the single geometric expression of three-dimensional point cloud data obtained through visual measurement to multi-reference structure expression by deducing the reference structure candidate set based on the assembling constraint set and generating the assembling path accessibility table, and improves the forward-looking nature and engineering guidance value of the assembling quality evaluation.

Description

Technical Field

[0001] This invention relates to the field of visual measurement technology, and in particular to a method for evaluating the quality of segment assembly based on three-dimensional point clouds. Background Technology

[0002] With the rapid development of urban rail transit and underground integrated utility tunnel projects, the assembly quality of precast tunnel segments in shield tunneling directly affects the overall stability and operational safety of the tunnel structure. Traditional methods for inspecting the assembly quality of tunnel segments mainly rely on manual measurement, total station measurement, or geometric error analysis based on a small number of key points. The inspection targets are mostly concentrated on single indicators such as joint width, misalignment, and circumferential closure error. In recent years, with the development of laser scanning technology, 3D reconstruction technology, and visual measurement technology, tunnel segment assembly inspection methods based on 3D point clouds have gradually become a research hotspot. By acquiring high-density spatial point data through multi-source visual measurement equipment, a comprehensive representation of the tunnel segment assembly status can be achieved.

[0003] Existing technologies for assessing the quality of tunnel segment assembly based on 3D point clouds still have certain shortcomings. Current methods typically rely on a single reference structure to align and assess the error of point cloud data, lacking a multi-reference structure competition mechanism. This results in a strong dependence on the initial reference structure, making it prone to misjudgment when local errors or data deviations exist, and failing to accurately reflect the true assembly status. Existing technologies mostly evaluate based on local geometric error indicators, failing to effectively incorporate structural constraint information such as assembly path, structural closure relationship, and joint continuity. They cannot dynamically analyze the accessibility and potential structural risks during the assembly process, resulting in assessment results that only remain at the static error level, making it difficult to support decision optimization and risk warning during construction. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for evaluating the quality of segment assembly based on three-dimensional point clouds to solve the problem of relying on a single reference structure and lacking comprehensive analysis of assembly path and structural constraints.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a method for evaluating the assembly quality of pipe segments based on three-dimensional point clouds. The method includes: collecting three-dimensional point cloud data of each pipe segment within the assembly area to be evaluated, and performing spatiotemporal consistency processing to form assembly point cloud data; extracting the outer pipe geometric contour and connection boundary features from the assembly point cloud data to form a pipe segment structural feature set; analyzing the circumferential and longitudinal connection relationships between pipe segments in the pipe segment structural feature set to generate a pipe segment connection relationship table; extracting the boundary morphology and seam distribution of the pipe segment structural feature set to obtain an assembly constraint set; deducing the reference assembly state of the pipe segment connection relationship table based on the assembly constraint set to generate a reference structure candidate set; comparing the consistency of the assembly structure between the reference structure candidate set and the assembly point cloud data to obtain the optimal assembly scheme; associating and mapping the boundary positions of each pipe segment in the optimal assembly scheme with the corresponding boundary positions in the assembly point cloud data to generate a pipe segment position offset relationship; and simulating the splicing of the optimal assembly scheme based on the pipe segment position offset relationship to generate an assembly path reachability table; evaluating the assembly quality of the assembly path reachability table and classifying the quality levels to generate an assembly quality evaluation table.

[0007] As a preferred embodiment of the segment assembly quality assessment method based on three-dimensional point clouds described in this invention, the specific steps for forming the assembly point cloud data are as follows: Perform time alignment, noise removal, and outlier filtering on the 3D point cloud data to form a cleaned point cloud sequence; Point cloud data from different perspectives in the purified point cloud sequence are aligned in space and mapped to the same reference coordinate system to form assembled point cloud data.

[0008] As a preferred embodiment of the segment assembly quality assessment method based on three-dimensional point clouds described in this invention, the specific steps for forming the segment structure feature set are as follows: By statistically analyzing the point cloud curvature distribution of the assembled point cloud data, the arc-shaped surface regions corresponding to each segment are extracted to form a set of segment surface distributions. Based on the distribution set of tunnel segment surfaces, the joint boundaries and boundary extension directions between each tunnel segment are extracted, and the boundary connection relationship between tunnel segments is established to form a tunnel segment structural feature set.

[0009] As a preferred embodiment of the segment assembly quality assessment method based on three-dimensional point clouds described in this invention, the specific steps for generating the segment connection relationship table are as follows: The segments on both sides of each joint boundary in the segment structure feature set are correlated and classified by direction. The correspondence is divided into circumferential connection relationship and longitudinal connection relationship, forming a connection relationship classification set. Based on the circumferential connection relationship, closed-path organization processing is performed on each segment to identify the circumferential arrangement order of each segment and generate a segment connection relationship table.

[0010] As a preferred embodiment of the segment assembly quality assessment method based on three-dimensional point clouds described in this invention, the specific steps for obtaining the assembly constraint set are as follows: Calculate the degree of boundary fit between adjacent segments in the segment structure feature set to form a boundary fit feature set; Based on the connection relationship classification set, the arrangement and closure of each segment are detected to form a closure feature set; The distribution continuity of each seam boundary is statistically analyzed, and then jointly organized with the boundary fitting feature set and the closure feature set to form an assembly constraint set.

[0011] As a preferred embodiment of the segment assembly quality assessment method based on three-dimensional point clouds described in this invention, the specific steps for generating the reference structure candidate set are as follows: Based on the segment connection relationship table, each segment is organized according to the connection relationship of the joint boundary to form an assembly arrangement sequence; Based on the assembly constraint set, the boundary alignment adjustment and connection position reorganization of the assembly arrangement sequence are performed to form an assembly structure sequence; By arranging the assembled structures in space, the sequence of assembled structures is transformed into a corresponding reference assembled state, forming a candidate set of reference structures.

[0012] As a preferred embodiment of the segment assembly quality assessment method based on three-dimensional point clouds described in this invention, the specific steps for obtaining the optimal assembly scheme are as follows: The reference assembly states in the candidate reference structure set are matched with the assembly point cloud data to establish the boundary correspondence between each reference assembly state and the assembly point cloud data, forming a set of matching relationships; Based on the matching relationship set, the consistency of the assembly structure of each reference assembly state is analyzed, sorted and filtered to obtain the optimal assembly scheme.

[0013] As a preferred embodiment of the segment assembly quality assessment method based on three-dimensional point clouds described in this invention, the specific steps for generating the segment position offset relationship are as follows: The boundary positions of each segment in the optimal assembly scheme are associated and matched with the corresponding boundary positions in the assembly point cloud data to establish a one-to-one correspondence between the boundaries of each segment, forming a boundary mapping set. The boundary position offsets of each segment in the boundary mapping set are decomposed in direction and calculated in magnitude to form a boundary displacement expression set; Based on the segment connection relationship table, the displacements of adjacent segments in the boundary displacement expression set are adjusted in a coordinated manner to form the segment position offset relationship.

[0014] As a preferred embodiment of the segment assembly quality assessment method based on three-dimensional point clouds described in this invention, the specific steps for generating the assembly path reachability table are as follows: Based on the optimal assembly scheme and the offset relationship of the segments, the segments are assembled step by step according to the connection order in the segment connection relationship table to form an assembly process sequence; In the splicing process sequence, spatial detection is performed on the positional relationship of adjacent segments after each splicing step to identify whether there is boundary interference, connection misalignment or local closure restriction during the splicing process, and to form an assembly path reachability table.

[0015] As a preferred embodiment of the segment assembly quality assessment method based on three-dimensional point clouds described in this invention, the specific steps for generating the assembly quality assessment table are as follows: By statistically analyzing the cumulative offset and connection status of the segments in each splicing path in the assembly path reachability table, we can identify the misalignment accumulation area and structural abnormal area during the assembly process, and form an abnormal area marker set. Based on the assembly path reachability table and the abnormal area marker set, the quality level of each segment's assembly status is classified, and an assembly quality assessment table is generated.

[0016] The beneficial effects of this invention are as follows: By extrapolating the candidate set of reference structures based on the assembly constraint set, the three-dimensional point cloud data obtained by visual measurement is upgraded from a single geometric expression to a multi-reference structure expression, realizing multi-angle structural reconstruction and comparative analysis of the segment assembly state, and improving the accuracy and stability of assembly state identification; by simulating the splicing based on the segment position offset relationship and generating an assembly path reachability table, the transformation from static visual measurement results to dynamic assembly process extrapolation is realized, so that the evaluation process not only reflects the current geometric deviation, but also can characterize the structural evolution and path feasibility during the assembly process, improving the foresight and engineering guidance value of the assembly quality assessment. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a method for assessing the quality of tunnel segment assembly based on 3D point clouds.

[0019] Figure 2 A schematic diagram illustrating the construction of point cloud structural features for the pipe segment.

[0020] Figure 3 This is a schematic diagram illustrating the assembly constraint derivation and optimal structure generation.

[0021] Figure 4 This is a schematic diagram for assembly path evaluation and quality judgment. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4 This is one embodiment of the present invention, which provides a method for evaluating the assembly quality of tunnel segments based on three-dimensional point clouds, including the following steps: S1. Collect the three-dimensional point cloud data of each segment in the assembly area to be evaluated, and perform spatiotemporal consistency processing to form the assembly point cloud data.

[0026] S1.1 Perform time alignment, noise removal, and outlier filtering on the 3D point cloud data to form a cleaned point cloud sequence.

[0027] It should be noted that the 3D point cloud data includes point cloud data of the outer surface of the pipe segments, point cloud data of the seam boundary area, and point cloud data of the connection positions. Among them, the point cloud data of the outer surface of the pipe segments is obtained by scanning the outer arc surface of each pipe segment. During the scanning process, continuous sampling is performed along the arc surface of the pipe segment to form a dense set of surface points. The point cloud data of the seam boundary area is obtained by scanning the boundary tracking position of adjacent pipe segments, so that the boundary points on both sides of the seam are continuously distributed in space. The point cloud data of the connection positions is obtained by performing local fine scanning of the connection parts between pipe segments to improve the spatial resolution of the connection area. The 3D scanning devices deployed at different locations collect data in the corresponding areas, and simultaneously record the acquisition time and spatial pose information of each scanning position, thereby forming 3D point cloud data containing multi-view information.

[0028] Based on the acquisition time corresponding to each sampling point, the point cloud data of the outer surface of the pipe segment, the point cloud data of the splice boundary area, and the point cloud data of the connection location are sorted according to a unified time axis. A unified time benchmark is constructed starting from the minimum acquisition time, and the point cloud data of different acquisition locations are mapped to the corresponding time positions according to the time difference, so that the various types of point cloud data form a consistent time series in the time dimension. After completing the time alignment, a fixed number of neighboring points (such as 15 to 30 neighboring points, determined according to the point cloud density of the assembled point cloud data) are selected as the center of each sampling point. The average distance between the current sampling point and the neighboring points is calculated, and the average distance between all sampling points is calculated. The average distance values ​​are used to form an average distance sequence. Based on the average distance sequence, the overall average value and standard deviation are calculated. Sampling points whose average distance is greater than the sum of the overall average value and standard deviation are identified as noise points and removed. After noise removal, the number of neighboring points is counted within a fixed spatial radius (e.g., 5 mm to 20 mm, determined according to the sampling interval of the assembled point cloud data) around each sampling point to form a neighboring point number sequence. The average value of the neighboring point number sequence is calculated. Sampling points whose number of neighboring points is less than the average value are identified as outliers and removed, thus obtaining a clean point cloud sequence that is consistent in time and continuous in spatial distribution.

[0029] S1.2 Align the point cloud data from different perspectives in the purified point cloud sequence in spatial position and map them to the same reference coordinate system to form assembled point cloud data.

[0030] It should be noted that, based on the spatial pose information corresponding to each sampling point, initial spatial alignment is performed on the point cloud data in the clean point cloud sequence to ensure that the data from different sampling locations are under a unified spatial reference. After the initial alignment is completed, point cloud data with overlapping spatial locations in the clean point cloud sequence are extracted, and the spatial distance between each sampling point in the overlapping area is calculated. By iteratively adjusting the spatial transformation relationship between the point cloud data, the overall spatial distance in the overlapping area is reduced and tends to stabilize. After the spatial alignment is completed, a unified reference coordinate system is selected as the spatial reference, and the point cloud data in the clean point cloud sequence is mapped to the reference coordinate system according to the corresponding spatial transformation relationship. The mapped point cloud data is then uniformly organized to form assembled point cloud data.

[0031] It should also be noted that when performing spatial alignment on point cloud data within overlapping areas, the point cloud data regions with overlapping spatial positions are identified in the cleaned point cloud sequence, and the correspondence between each sampling point within the overlapping area is established. After establishing the correspondence, the point cloud data in the cleaned point cloud sequence is adjusted overall based on the spatial position differences between the corresponding sampling points to reduce the spatial distance between the corresponding sampling points. After each position adjustment, the spatial distance between the corresponding sampling points is recalculated, and the subsequent position adjustment range is gradually reduced based on the sequence of changes in spatial distance during multiple consecutive position adjustments until the change in spatial distance remains consistent in multiple consecutive adjustments (e.g., more than 3 times), thus completing the spatial alignment process.

[0032] S2. Extract the geometric contour and connection boundary features of the outer pipe from the assembled point cloud data to form a pipe segment structure feature set. Analyze the circumferential and longitudinal connection relationships between the pipe segments in the pipe segment structure feature set to generate a pipe segment connection relationship table.

[0033] S2.1 By statistically analyzing the point cloud curvature distribution of the assembled point cloud data, extract the corresponding arc-shaped surface area of ​​each segment to form a set of segment surface distributions.

[0034] It should be noted that in assembling the point cloud data, a set of sampling points within the neighborhood of each sampling point is selected, and the curvature values ​​of the corresponding sampling points are statistically analyzed based on the spatial positional relationship of each sampling point within the neighborhood, forming a point cloud curvature sequence. The point cloud curvature sequence is sorted according to the magnitude of the curvature values, and the curvature difference between adjacent sampling points after sorting is obtained, forming a curvature difference sequence. In the curvature difference sequence, each curvature difference is sorted according to the magnitude of its value, and the change in the difference between adjacent curvature differences is statistically analyzed, forming a difference change sequence. The sequence of difference changes is used to identify the position with the largest difference change as the curvature segment position. Based on the curvature segment position, the assembled point cloud data is divided into regions to form multiple continuous curvature regions. Spatial connectivity detection is performed on the sampling points in each continuous curvature region, and the number of connected points corresponding to each continuous curvature region is counted to form a connected point number sequence. In the connected point number sequence, the continuous curvature regions with a number of connected points greater than the average number of connected points are selected as the arc surface regions of the tunnel segment and marked to form a set of tunnel segment surface distributions.

[0035] It should also be noted that spatial connectivity detection expands point by point within each curvature continuous region, starting from a sampling point, according to the spatial adjacency relationship between sampling points. Sampling points that can reach each other through adjacency are grouped into the same connected set, and the number of sampling points in each connected set is counted to form a sequence of connected point counts.

[0036] S2.2 Based on the distribution set of tunnel segment surfaces, extract the joint boundaries and boundary extension directions between each tunnel segment and establish the boundary connection relationship between tunnel segments to form a tunnel segment structure feature set.

[0037] It should be noted that, in the distribution set of tunnel segment surfaces, an adjacency region search is performed on any two adjacent surface regions to obtain sampling point pairs where the spatial distance between the two surface regions is less than the average spacing between adjacent surface regions. The spatial positions corresponding to the sampling point pairs are used as candidate points for the seam boundary, forming a set of candidate seam boundary points. The sampling points in the set of candidate seam boundary points are sorted according to their spatial positions, and a continuous boundary point sequence is constructed based on the spatial distance between adjacent sampling points. This continuous boundary point sequence is used as the seam boundary. The direction vectors between adjacent sampling points in the seam boundary are counted and normalized to form a boundary extension direction sequence. Based on the two surface regions corresponding to the seam boundary, a one-to-one correspondence is established between the tunnel segments corresponding to the two surface regions. The connection direction between the tunnel segments is marked by combining the boundary extension direction sequence to form a tunnel segment structural feature set.

[0038] S2.3. Correspondingly associate and classify the segments on both sides of each joint boundary in the segment structure feature set, and divide the correspondence into circumferential connection relationship and longitudinal connection relationship to form a connection relationship classification set.

[0039] It should be noted that, in the segment structure feature set, the boundary extension direction sequence corresponding to each joint boundary is read, and the two segments corresponding to each joint boundary are established with a one-to-one segment association relationship according to the same joint boundary, forming a segment correspondence set; the direction vectors in the boundary extension direction sequence are normalized and vector summation is performed to obtain the direction accumulation vector, the direction accumulation vector is normalized to obtain the main direction vector, and the main direction vector is projected onto the overall spatial distribution direction of the assembled point cloud data, and the direction angle between the main direction vector and each direction component is calculated to form a direction angle sequence; the direction angle sequence is sorted, and the difference between adjacent direction angles is calculated to form a direction difference sequence, the corresponding position with the largest difference in the direction difference sequence is extracted as the direction classification boundary position, and the segment correspondence set is divided into two categories based on the boundary position, one category of segment correspondences corresponding to the main direction distributed along the circumferential direction is regarded as the circumferential connection relationship, and the other category of segment correspondences corresponding to the main direction distributed along the axial direction is regarded as the longitudinal connection relationship, thus forming a connection relationship classification set.

[0040] It should also be noted that the changes in the spatial coordinates of each sampling point in different directions are statistically analyzed, and the direction with the largest change is selected as the overall spatial distribution direction.

[0041] S2.4. Perform closed-path organization processing on each segment according to the circumferential connection relationship, identify the arrangement order of each segment in the circumferential direction, and generate a segment connection relationship table.

[0042] It should be noted that the two segments corresponding to each circumferential connection are arranged sequentially as adjacent segments to form a segment adjacency sequence. Any segment in the segment adjacency sequence is selected as the starting segment, and the next adjacent segment corresponding to the current segment is found sequentially according to the circumferential connection. This next adjacent segment is added to the permutation sequence, and the set of segments already added to the permutation sequence is recorded, forming the permutation sequence. During the permutation sequence construction process, if the currently found next adjacent segment already exists in the set of segments already added to the permutation sequence, the current permutation sequence is determined as a closed permutation sequence, and the segments in the closed permutation sequence are determined as the circumferential permutation order according to the order of addition. All circumferential permutation orders corresponding to closed permutation sequences are summarized and organized to form a segment connection relationship table.

[0043] S3. Extract the boundary morphology and joint distribution of the segment structure feature set, obtain the assembly constraint set, deduce the reference assembly state of the segment connection relationship table based on the assembly constraint set, and generate a reference structure candidate set.

[0044] S3.1 Calculate the degree of boundary fit between adjacent segments in the segment structure feature set to form a boundary fit feature set.

[0045] It should be noted that the segment identifier, joint boundary identifier, and corresponding boundary fit degree of each group of adjacent segments are associated and recorded to form a boundary fit record item; all boundary fit record items are summarized and organized according to the arrangement order of the joint boundary in the segment structure feature set, so that each joint boundary corresponds to a unique boundary fit record item, forming a boundary fit feature set.

[0046] The expression for calculating the degree of boundary fit is: ; in, Indicates the first The first segment and the first The degree of boundary fit between individual pipe segments is a dimensionless quantity with a value range of [0, 1]. This indicates the number of sampling points involved in the calculation within the seam boundary. Indicates the first The corresponding distance between two adjacent tunnel segment boundaries for each sampling point This represents the average value of all corresponding distances within the seam boundary.

[0047] S3.2. Based on the connection relationship classification set, detect the arrangement and closure of each segment to form a closure feature set.

[0048] It should be noted that the circumferential arrangement order of each segment is read from the segment connection relationship table, and the connection between the first and last segments is checked for closure based on the circumferential connection relationship between adjacent segments in the circumferential arrangement order; the number of circumferential connection relationships corresponding to each segment in the circumferential arrangement order is counted to form a connection count sequence; the arrangement order in which the connection count of each segment in the connection count sequence is two and there is a circumferential connection relationship between the first and last segments is marked as a closed arrangement state, and the arrangement order that does not meet the closed arrangement state is marked as a non-closed arrangement state; the arrangement order position and corresponding closure state of each segment are associated and recorded to form a closure feature set.

[0049] S3.3 Statistically analyze the distribution continuity of each seam boundary, and combine it with the boundary fitting feature set and the closure feature set to form an assembly constraint set.

[0050] It should be noted that, in the segment structure feature set, the continuous boundary point sequence corresponding to each joint boundary is read, and the spacing change between adjacent sampling points is counted according to the spatial arrangement order of the sampling points in the continuous boundary point sequence, forming a spacing change sequence. The spacing change sequence is sorted according to the value size rule, and the difference between adjacent spacing changes is calculated, forming a change difference sequence. The position corresponding to the largest difference in the change difference sequence is determined as the continuous segment position, and the continuous boundary point sequence is segmented based on the continuous segment position. The number of sampling points corresponding to each segment is counted to form a continuous segment length sequence. The continuous segment length is greater than the average of the continuous segment length sequences. The segmented values ​​are marked as continuous seam boundaries, thus obtaining the distribution continuity results of the seam boundaries. The distribution continuity results corresponding to each seam boundary are associated with the boundary fitting record items in the boundary fitting feature set and the closure state in the closure feature set. Based on the vertical connection relationship, the distribution continuity results of the corresponding seam boundaries between adjacent rings are consistent and aligned. The seam boundaries with consistent continuous boundary positions in the vertical connection relationship are marked as continuous assembly boundaries, and the seam boundaries with inconsistent positions are marked as discontinuous assembly boundaries. The boundary fitting record items, closure states and distribution continuity results are uniformly organized to form an assembly constraint set.

[0051] S3.4 Based on the segment connection relationship table, organize each segment according to the connection relationship of the joint boundary to form an assembly arrangement sequence.

[0052] It should be noted that the circumferential arrangement order recorded in the segment connection relationship table is used as the basic order. Each segment in the circumferential arrangement order is unfolded according to the connection relationship corresponding to the joint boundary to form a basic assembly sequence. In the basic assembly sequence, the corresponding segments between adjacent rings are associated with the longitudinal connection relationship. The segments in the same longitudinal connection chain are organized hierarchically according to the connection order to form an assembly arrangement sequence that includes the circumferential order and the longitudinal correspondence. The arrangement position of each segment in the assembly arrangement sequence is sequentially numbered, and the joint boundary connection relationship and the longitudinal correspondence relationship corresponding to each segment are recorded synchronously to form the assembly arrangement sequence.

[0053] S3.5. Based on the assembly constraint set, the boundary alignment adjustment and connection position reorganization of the assembly arrangement sequence are performed to form an assembly structure sequence.

[0054] It should be noted that the connection relationship of the joint boundary of each segment is read in the assembly arrangement sequence, and the consistency of the boundary position between adjacent segments in the assembly arrangement sequence is checked by combining the boundary fitting record, closure state and distribution continuity result of the corresponding joint boundary in the assembly constraint set. For adjacent segments that do not meet the boundary fitting constraint, the boundary position of the corresponding segment is corrected according to the boundary fitting degree in the boundary fitting record. For joint boundaries that do not meet the distribution continuity constraint, the segmented position of the continuous boundary point sequence is adjusted according to the distribution continuity result to keep the change in boundary spacing between adjacent segments consistent. For the assembly arrangement sequence that does not meet the closure state, the connection position between the first and last segments is repositioned according to the closure state recorded in the closure feature set to form a circumferential connection between the first and last segments. The spatial position of each segment in the assembly arrangement sequence and the corresponding joint boundary connection relationship are uniformly organized to form the assembly structure sequence.

[0055] It should also be noted that when performing boundary alignment adjustment on adjacent segments that do not meet the boundary fitting constraints, the boundary position of the adjacent segments is successively corrected based on the spatial distance between the corresponding sampling points in the boundary fitting record. After each position correction, the average spatial distance between the corresponding sampling points is calculated to form a spatial distance sequence. The difference between the average spatial distances in two consecutive position corrections is calculated to form a difference change sequence. When the current difference in the difference change sequence is less than the average value of the difference change sequence, the position correction process is stopped.

[0056] S3.6. The assembly structure sequence is converted into a corresponding reference assembly state through spatial arrangement to form a candidate set of reference structures.

[0057] It should be noted that the arrangement order of each segment and the corresponding joint boundary connection relationship are read in the assembly structure sequence. Based on the spatial positional relationship of the segments that satisfy the assembly constraint set in the assembly structure sequence, the spatial position of each segment is expressed according to the arrangement order, so that each segment establishes a corresponding positional relationship in space according to the joint boundary connection relationship, forming a spatial layout structure. The spatial position and joint boundary connection relationship of each segment in the spatial layout structure are uniformly recorded to form a reference assembly state. The reference assembly states corresponding to each spatial layout structure are summarized and organized to form a reference structure candidate set.

[0058] S4. Compare the consistency of the assembly structure between the candidate reference structure set and the assembly point cloud data to obtain the optimal assembly scheme. Associate and map the boundary positions of each segment in the optimal assembly scheme with the corresponding boundary positions in the assembly point cloud data to generate the segment position offset relationship.

[0059] S4.1 Match each reference assembly state in the candidate set of reference structures with the assembly point cloud data, establish the boundary correspondence between each reference assembly state and the assembly point cloud data, and form a set of matching relationships.

[0060] It should be noted that the boundary positions of each segment in each reference assembly state are read in the reference structure candidate set, and the corresponding boundary positions of the segments are read in the assembly point cloud data. A one-to-one correspondence between the reference assembly state and the assembly point cloud data is established according to the segment identifier. In each group of corresponding segment relationships, the boundary positions in the reference assembly state and the corresponding boundary positions in the assembly point cloud data are synchronously sorted along the extension direction of the seam boundary, so that the sampling points in the same boundary order position form corresponding sampling point pairs, and a boundary corresponding point sequence is obtained. The corresponding sampling point pairs in the boundary corresponding point sequence are associated and recorded, and all boundary corresponding point sequences corresponding to each reference assembly state are summarized and organized to form a matching relationship set.

[0061] S4.2 Based on the matching relationship set, analyze the consistency of the assembly structure of each reference assembly state, sort and filter them to obtain the optimal assembly scheme.

[0062] It should be noted that, in the matching relationship set, all boundary corresponding point sequences corresponding to each reference assembly state are read, and the boundary fitting degree corresponding to each seam boundary is calculated based on the spatial distance between each corresponding sampling point in the boundary corresponding point sequence; the boundary fitting degrees corresponding to all seam boundaries in the same reference assembly state are summarized to form a boundary fitting degree sequence; the number of seam boundaries with a boundary fitting degree greater than the average value of the current boundary fitting degree sequence is counted in the boundary fitting degree sequence, and its proportion in all seam boundaries is calculated as the fitting consistency value of the current reference assembly state; the closure state in the closure feature set corresponding to the current reference assembly state and the distribution continuity result corresponding to the assembly constraint set are read, and the proportion of the number of seam boundaries that satisfy the closure state and have consistent distribution continuity is used as the structural constraint consistency value; based on the fitting consistency value and the structural constraint consistency value, each reference assembly state is sorted, and the reference assembly state with the highest consistency value in the sorting result is selected as the optimal assembly scheme.

[0063] It should also be noted that when sorting and filtering the reference assembly states, the reference assembly states are filtered based on the structural constraint consistency value, and the set of reference assembly states with the largest structural constraint consistency value is retained; in the set of reference assembly states, the reference assembly state with the largest fitting consistency value is selected as the optimal assembly scheme.

[0064] S4.3. Associate and match the boundary positions of each segment in the optimal assembly scheme with the corresponding boundary positions in the assembly point cloud data to establish a one-to-one correspondence between the boundaries of each segment and form a boundary mapping set.

[0065] It should be noted that the joint boundary positions of each segment are read from the optimal assembly scheme and the joint boundary positions of the corresponding segments are read from the assembly point cloud data. The segment correspondence between the optimal assembly scheme and the assembly point cloud data is established based on the segment identifier. In each group of corresponding segment relationships, the boundary positions in the optimal assembly scheme and the corresponding boundary positions in the assembly point cloud data are synchronously sorted along the extension direction of the joint boundary, so that the sampling points in the same boundary order position form corresponding sampling point pairs, and a boundary corresponding point sequence is obtained. Each corresponding sampling point pair in the boundary corresponding point sequence is associated and recorded one by one, and all the boundary corresponding point sequences corresponding to each segment are summarized and organized to form a boundary mapping set.

[0066] S4.4 Perform directional decomposition and amplitude calculation on the boundary position offset of each segment in the boundary mapping set to form a boundary displacement expression set.

[0067] It should be noted that the boundary corresponding point sequence of each segment is read from the boundary mapping set, and the spatial position difference vector of each pair of corresponding sampling points is obtained to form a boundary displacement vector sequence. Based on the boundary extension direction sequence of the joint boundary of each pair of corresponding sampling points, the component along the boundary extension direction is extracted from the boundary displacement vector as the tangential displacement component. Based on the spatial position relationship of the boundary corresponding points, a direction perpendicular to the boundary extension direction is constructed, and the component of the boundary displacement vector in the perpendicular direction is taken as the normal displacement component to form a displacement component sequence after directional decomposition. The vector length of each displacement vector in the displacement component sequence is counted to form a displacement amplitude sequence. The tangential displacement component, normal displacement component and displacement amplitude of each segment are associated and recorded to form a boundary displacement expression set.

[0068] S4.5 Based on the segment connection relationship table, adjust the displacement of adjacent segments in the boundary displacement expression set to form the segment position offset relationship.

[0069] It should be noted that the circumferential and longitudinal connection relationships between each segment are read from the segment connection relationship table, and the tangential displacement components, normal displacement components, and displacement amplitudes corresponding to the circumferential and longitudinal connection relationships are read from the boundary displacement expression set. The displacement expressions corresponding to each group of adjacent segments are correlated to obtain the differences in tangential displacement components, normal displacement components, and displacement amplitudes on both sides of the joint boundary between adjacent segments, forming an adjacent displacement difference sequence. The adjacent displacement difference sequence is arranged according to the segment connection order, and the changes between adjacent displacement differences are statistically analyzed. A displacement change sequence is generated. The displacement expressions of adjacent segments whose displacement changes are greater than the average value of the displacement change sequence are marked as displacement expressions to be adjusted, and the displacement expressions of adjacent segments whose displacement changes are not greater than the average value of the displacement change sequence are marked as displacement expressions to be maintained. The displacement expressions to be adjusted are synchronously corrected according to the corresponding adjacent segment's maintained displacement expression, so that adjacent segments in the same connection chain maintain a continuous change relationship in tangential displacement component, normal displacement component, and displacement amplitude. The corrected displacement expressions of each segment are summarized and organized according to the segment identifier to form the segment position offset relationship.

[0070] S5. Based on the segment position offset relationship, perform splicing simulation on the optimal assembly scheme, generate an assembly path reachability table, evaluate the assembly quality of the assembly path reachability table, classify the quality level, and generate an assembly quality evaluation table.

[0071] S5.1 Based on the optimal assembly scheme and the segment position offset relationship, each segment is assembled step by step according to the connection order in the segment connection relationship table to form an assembly process sequence.

[0072] It should be noted that, in the optimal assembly scheme, the spatial position and joint boundary connection relationship of each segment are read, and the tangential displacement component, normal displacement component, and displacement amplitude of each segment are read from the segment position offset relationship. The tangential displacement component, normal displacement component, and displacement amplitude are loaded to the spatial position of the corresponding segment in the optimal assembly scheme to form the actual splicing position of each segment. According to the connection order in the segment connection relationship table, the actual splicing position of the starting segment is used as the splicing starting point. The next segment that has a connection relationship with the starting segment is loaded to the corresponding actual splicing position. According to the joint boundary connection relationship, the loaded segment is connected to the next segment to form the current splicing sequence. The actual splicing positions of subsequent segments are continuously loaded along the connection order in the segment connection relationship table, and the current splicing sequence formed after each loading is associated and recorded according to the splicing order to form the splicing process sequence.

[0073] S5.2 In the splicing process sequence, spatial detection is performed on the positional relationship of adjacent segments after each splicing step to identify whether there is boundary interference, connection misalignment or local closure restriction during the splicing process, and to form an assembly path reachability table.

[0074] It should be noted that in the splicing process sequence, the current splicing sequence after each splicing step is read according to the splicing order. The boundary position relationship between the newly added segment and the adjacent spliced ​​segments is read from the current splicing sequence. The normal spacing sequence, tangential displacement difference sequence, and closed spacing sequence between the first and last segments are obtained from the corresponding sampling point pairs of adjacent segments. Negative sampling point pairs in the normal spacing sequence are counted, and splicing states with negative sampling point pairs are marked as boundary interference states. The tangential displacement difference sequence is sorted, and the change between adjacent tangential displacement differences is calculated to form a misalignment change sequence. The splicing state where the amount of change is greater than the average value of the misalignment change sequence is marked as a connection misalignment state. The splicing state where there is no connection between the first and last segments and the minimum closure spacing in the closure spacing sequence is greater than the average value of the closure spacing sequence is marked as a locally restricted closure state. The boundary interference state, connection misalignment state and locally restricted closure state corresponding to each splicing step are associated and recorded according to the splicing order. The splicing path without boundary interference state, connection misalignment state and locally restricted closure state is marked as an reachable splicing path. The splicing path with any state is marked as a restricted splicing path, forming an assembly path reachability table.

[0075] S5.3 By statistically analyzing the cumulative offset and connection status of segments in each splicing path in the assembly path reachability table, identify the misalignment accumulation area and structural abnormal area that exist during the assembly process, and form an abnormal area marker set.

[0076] It should be noted that the splicing process sequence and corresponding path category of each splicing path are read from the splicing path reachability table, and the segment position offset relationship of the corresponding segments in each splicing path is read. The normal displacement components of each segment are accumulated according to the splicing order to form a normal offset accumulation sequence, and the offset difference between adjacent segments in the normal offset accumulation sequence is calculated to form an offset difference sequence. The offset difference sequence is sorted according to the value size rule and the change between adjacent offset differences is calculated to form an offset change sequence. The position with the largest change in the offset change sequence is determined as the misalignment segment position, and the segment at the corresponding position is marked as the misalignment accumulation area. After reading the path category of the splicing path, the segment position in the splicing path belonging to the restricted splicing path is directly marked as the structural abnormal position, and the segment position in the reachable splicing path with boundary interference state, connection misalignment state, or local closure restricted state is marked as the local abnormal position. The misalignment accumulation area, structural abnormal position, and local abnormal position are uniformly organized and associated with the segment identification to form an abnormal area marking set.

[0077] S5.4 Based on the assembly path reachability table and abnormal area marker set, classify the quality level of each segment's assembly status and generate an assembly quality assessment table.

[0078] It should be noted that the path category corresponding to each splicing path and the splicing status mark of each segment in the splicing process sequence are read from the assembly path reachability table. The misalignment accumulation area mark, structural anomaly location mark, and local anomaly location mark corresponding to each segment are read from the anomaly area mark set. The number of times each segment is marked as a structural anomaly location and the number of times it is marked as a local anomaly location in all splicing paths are counted to form an anomaly mark statistical sequence. Segments with a structural anomaly location mark count greater than zero in the anomaly mark statistical sequence are marked as high-risk segments. Segments not marked as structural anomaly locations but with a local anomaly location mark count greater than the average of the anomaly mark statistical sequence are marked as key correction segments. Segments not marked as structural anomaly locations but with a local anomaly location mark count not greater than the average of the anomaly mark statistical sequence are marked as stable assembly segments. The quality level marking results of each segment, the corresponding anomaly mark type, and their positions in the splicing path are associated and recorded to form an assembly quality assessment table.

[0079] In summary, this invention improves the accuracy and stability of assembly status identification by: firstly, elevating the 3D point cloud data acquired through visual measurement from a single geometric representation to a multi-reference structural representation based on the assembly constraint set to deduce the candidate reference structure set; secondly, enabling multi-angle structural reconstruction and comparative analysis of the segment assembly status by deduce the candidate reference structure set based on the segment position offset relationship, and by generating an assembly path reachability table, transforming static visual measurement results into dynamic assembly process deduction. This allows the evaluation process to not only reflect the current geometric deviation but also characterize the structural evolution and path feasibility during assembly, enhancing the foresight and engineering guidance value of assembly quality assessment.

[0080] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for evaluating the assembly quality of tunnel segments based on three-dimensional point clouds, characterized in that, include: Collect three-dimensional point cloud data of each segment in the assembly area to be evaluated, and perform spatiotemporal consistency processing to form assembly point cloud data; Extract the geometric contour and connection boundary features of the outer tube from the assembled point cloud data to form a segment structure feature set. Analyze the circumferential and longitudinal connection relationships between segments in the segment structure feature set to generate a segment connection relationship table. Extract the boundary morphology and joint distribution of the segment structure feature set, obtain the assembly constraint set, deduce the reference assembly state of the segment connection relationship table based on the assembly constraint set, and generate a reference structure candidate set; The consistency of the assembled structure between the candidate reference structure set and the assembled point cloud data is compared to obtain the optimal assembly scheme. The boundary positions of each segment in the optimal assembly scheme are associated and mapped with the corresponding boundary positions in the assembled point cloud data to generate the segment position offset relationship. Based on the segment position offset relationship, the optimal assembly scheme is simulated to generate an assembly path accessibility table, the assembly quality of the assembly path accessibility table is evaluated, the quality level is divided, and an assembly quality evaluation table is generated.

2. The method for evaluating the assembly quality of tunnel segments based on three-dimensional point clouds as described in claim 1, characterized in that, The specific steps for forming the assembled point cloud data are as follows: Perform time alignment, noise removal, and outlier filtering on the 3D point cloud data to form a cleaned point cloud sequence; Point cloud data from different perspectives in the purified point cloud sequence are aligned in space and mapped to the same reference coordinate system to form assembled point cloud data.

3. The method for evaluating the assembly quality of tunnel segments based on three-dimensional point clouds as described in claim 2, characterized in that, The specific steps for forming the segment structure feature set are as follows: By statistically analyzing the point cloud curvature distribution of the assembled point cloud data, the arc-shaped surface regions corresponding to each segment are extracted to form a set of segment surface distributions. Based on the distribution set of tunnel segment surfaces, the joint boundaries and boundary extension directions between each tunnel segment are extracted, and the boundary connection relationship between tunnel segments is established to form a tunnel segment structural feature set.

4. The method for evaluating the quality of tunnel segment assembly based on three-dimensional point clouds as described in claim 3, characterized in that, The specific steps for generating the segment connection relationship table are as follows: The segments on both sides of each joint boundary in the segment structure feature set are correlated and classified by direction. The correspondence is divided into circumferential connection relationship and longitudinal connection relationship, forming a connection relationship classification set. Based on the circumferential connection relationship, closed-path organization processing is performed on each segment to identify the circumferential arrangement order of each segment and generate a segment connection relationship table.

5. The method for evaluating the assembly quality of tunnel segments based on three-dimensional point clouds as described in claim 4, characterized in that, The specific steps for obtaining the assembly constraint set are as follows: Calculate the degree of boundary fit between adjacent segments in the segment structure feature set to form a boundary fit feature set; Based on the connection relationship classification set, the arrangement and closure of each segment are detected to form a closure feature set; The distribution continuity of each seam boundary is statistically analyzed, and then jointly organized with the boundary fitting feature set and the closure feature set to form an assembly constraint set.

6. The method for evaluating the assembly quality of tunnel segments based on three-dimensional point clouds as described in claim 1, characterized in that, The specific steps for generating the candidate set of reference structures are as follows: Based on the segment connection relationship table, each segment is organized according to the connection relationship of the joint boundary to form an assembly arrangement sequence; Based on the assembly constraint set, the boundary alignment adjustment and connection position reorganization of the assembly arrangement sequence are performed to form an assembly structure sequence; By arranging the assembled structures in space, the sequence of assembled structures is transformed into a corresponding reference assembled state, forming a candidate set of reference structures.

7. The method for evaluating the assembly quality of tunnel segments based on three-dimensional point clouds as described in claim 2 or 6, characterized in that, The specific steps for obtaining the optimal assembly scheme are as follows: The reference assembly states in the candidate reference structure set are matched with the assembly point cloud data to establish the boundary correspondence between each reference assembly state and the assembly point cloud data, forming a set of matching relationships; Based on the matching relationship set, the consistency of the assembly structure of each reference assembly state is analyzed, sorted and filtered to obtain the optimal assembly scheme.

8. The method for evaluating the assembly quality of tunnel segments based on three-dimensional point clouds as described in claim 7, characterized in that, The specific steps for generating the segment position offset relationship are as follows: The boundary positions of each segment in the optimal assembly scheme are associated and matched with the corresponding boundary positions in the assembly point cloud data to establish a one-to-one correspondence between the boundaries of each segment, forming a boundary mapping set. The boundary position offsets of each segment in the boundary mapping set are decomposed in direction and calculated in magnitude to form a boundary displacement expression set; Based on the segment connection relationship table, the displacements of adjacent segments in the boundary displacement expression set are adjusted in a coordinated manner to form the segment position offset relationship.

9. The method for evaluating the assembly quality of tunnel segments based on three-dimensional point clouds as described in claim 8, characterized in that, The specific steps for generating the assembly path reachability table are as follows: Based on the optimal assembly scheme and the offset relationship of the segments, the segments are assembled step by step according to the connection order in the segment connection relationship table to form an assembly process sequence; In the splicing process sequence, spatial detection is performed on the positional relationship of adjacent segments after each splicing step to identify whether there is boundary interference, connection misalignment or local closure restriction during the splicing process, and to form an assembly path reachability table.

10. The method for evaluating the assembly quality of tunnel segments based on three-dimensional point clouds as described in claim 9, characterized in that, The specific steps for generating the assembly quality assessment form are as follows: By statistically analyzing the cumulative offset and connection status of the segments in each splicing path in the assembly path reachability table, we can identify the misalignment accumulation area and structural abnormal area during the assembly process, and form an abnormal area marker set. Based on the assembly path reachability table and the abnormal area marker set, the quality level of each segment's assembly status is classified, and an assembly quality assessment table is generated.

Images