A method for detecting continuous segment of shield tunnel segment joint deformation

Abstract

The application discloses a kind of shield tunnel segment joint deformation continuous section detection method.The method comprises the following steps: extracting target joint point cloud, and constructing local orthogonal reference coordinate system based on joint main direction and joint two sides segment surface normal vector;Minimum analysis unit is divided along the joint direction, and point cloud quality field is constructed according to point cloud density, local missing rate, two sides point number balance degree, normal dispersion and fitting residual;Based on quality field, the analysis window is expanded, merged or invalid determination is carried out, the amount of error and the amount of opening are calculated in effective window, and the results of overlapping window are fused according to window confidence degree;According to the fused deformation, identify the continuous section of error, opening anomaly and composite anomaly.

Description

Technical Field

[0001] This invention relates to the field of shield tunnel structure inspection and three-dimensional point cloud data processing technology, and in particular to a method for detecting continuous sections of shield tunnel segment joint deformation based on handheld laser scanning of non-uniform point clouds. Background Technology

[0002] Shield tunnels are typically assembled from multiple rings of prefabricated segments. Adjacent segments and adjacent lining rings are connected by bolts, waterstops, and other structural elements to form the tunnel lining structure. Due to factors such as construction disturbance, ground settlement, long-term operational loads, groundwater effects, and surrounding engineering projects, the joints of shield tunnel segments are prone to defects such as misalignment, opening, localized damage, and water leakage. Among these, the amount of misalignment and opening at the segment joints are important visual indicators reflecting the service condition of the shield tunnel structure, and are of great significance for evaluating tunnel operational safety, locating defects, and conducting maintenance.

[0003] Existing methods for detecting deformation at tunnel segment joints mainly include manual contact measurement, mechanical stylus measurement, mobile laser scanning measurement, fixed three-dimensional laser scanning measurement, and automatic identification methods based on images or point clouds. Manual measurement typically uses tools such as rulers, feeler gauges, and vernier calipers to measure local joint locations. While this method is simple, it suffers from low efficiency, high subjectivity, and difficulty in generating continuous data. Mechanical stylus or electrical measuring devices can improve the accuracy of measuring local joint misalignment and opening to some extent, but they are still contact measurement methods. Their efficiency and adaptability to the field are limited by tunnel space, joint location, operator posture, and equipment deployment conditions.

[0004] With the development of 3D laser scanning technology, point cloud-based methods for detecting joints in shield tunnel segments have been gradually applied. Existing technologies can acquire point clouds of shield tunnels using mobile laser scanning equipment, extract circumferential or longitudinal joint point clouds using point cloud recognition models, and then obtain the segment misalignment amount through methods such as local symmetrical point measurement, cross-section superposition, circular model fitting, or corresponding position distance calculation.

[0005] Some existing technologies can also identify the range of continuous misalignment in shield tunnels by statistically analyzing the start and end positions, continuous length, and average misalignment amount of the misalignment area based on measurement results that continuously exceed the misalignment threshold. These methods typically rely on a relatively complete tunnel cross-section point cloud, a relatively stable moving scan trajectory, and the fitting of symmetrical positions on either side of the cross-section center or joint.

[0006] However, the scanning trajectory and observation posture of handheld laser scanning equipment are easily affected by manual operation. Point clouds along the seam often suffer from uneven density distribution, local occlusion, unbalanced point counts on both sides of the seam, and fluctuations in local normals and fitting residuals. Under these circumstances, misalignment detection methods that rely on complete cross-sections, fitted center points, or fixed symmetrical positions are difficult to reliably determine the measurement benchmark. Fixed-length windows or ordinary sliding windows are also prone to outputting unreliable local deformation results in low-quality point cloud regions.

[0007] Existing technologies typically focus on detecting misalignment of tunnel segments, but have not yet established a point cloud quality evaluation mechanism that distributes the point cloud along the joint direction for handheld non-uniform point clouds. They also do not adaptively expand, merge, or remove analysis windows based on the point cloud quality at different locations. Furthermore, existing technologies lack the technical means to jointly calculate misalignment and opening within a unified local reference coordinate system of the joint, and to perform confidence fusion of overlapping measurement results based on the reliability of the window data.

[0008] Therefore, it is necessary to provide a method for detecting the joints of shield tunnel segments using handheld laser scanning of non-uniform point clouds. Without relying on the global tunnel centerline and the fitting center of the complete cross-section, the method adaptively generates an analysis window based on the quality differences of the point cloud along the joint, jointly obtains the misalignment amount and the opening amount, and stably identifies continuous sections of misalignment anomalies, opening anomalies, and misalignment-opening composite anomalies. Summary of the Invention

[0009] The objective of this invention is achieved through the following technical solutions.

[0010] The purpose of this invention is to provide a method for detecting continuous deformation sections of shield tunnel segment joints, in order to solve the problems of existing point cloud detection methods being susceptible to uneven point cloud density, local occlusion, attitude deviation, imbalance of the number of points on both sides of the joint, and local fitting errors under handheld scanning conditions.

[0011] To achieve the above objectives, the present invention provides a method for detecting continuous deformation sections of tunnel segment joints, comprising the following steps:

[0012] Extract the point cloud of the target joint and construct a local orthogonal reference coordinate system based on the main direction of the joint and the surface normal vectors of the segments on both sides of the joint;

[0013] The smallest analysis unit is divided along the main direction of the seam. A point cloud quality field is constructed based on the point cloud density, local missing rate, balance of points on both sides, normal dispersion and fitting residual. The analysis window is expanded, merged or invalidated based on the point cloud quality field.

[0014] Calculate the misalignment and opening within the effective window, and fuse the results of overlapping windows according to the window confidence level;

[0015] Identify continuous sections of misalignment anomalies, opening anomalies, and compound anomalies based on the fusion deformation.

[0016] Furthermore, the step of extracting the target joint point cloud and constructing a local orthogonal reference coordinate system based on the main direction of the joint and the surface normal vectors of the segments on both sides of the joint includes:

[0017] S1: Use a handheld laser scanning device to acquire three-dimensional point cloud data of the joint area of ​​shield tunnel segments;

[0018] S2: Perform outlier removal, voxel downsampling, normal vector estimation, and local smoothing on the three-dimensional point cloud data to obtain a standardized point cloud of the seam region;

[0019] S3: Extract candidate seam point clouds based on the local geometric abrupt change features of the standardized point cloud of the seam region, and perform connectivity screening on the candidate seam point clouds to obtain the target seam point cloud.

[0020] S4: Perform robust principal direction analysis on the target joint point cloud to obtain the joint principal direction; perform direction consistency correction and weighting on the average normal vectors of the segments on both sides of the joint, and orthogonalize the weighted direction vectors relative to the joint principal direction to obtain the misalignment direction axis; obtain the opening direction axis based on the cross product of the joint principal direction axis and the misalignment direction axis, thereby constructing a mutually orthogonal local reference coordinate system for the joint.

[0021] Furthermore, the step of dividing the analysis window into minimum analysis units along the main direction of the seam, constructing a point cloud quality field based on point cloud density, local missing rate, balance of points on both sides, normal dispersion, and fitting residual, and then expanding, merging, or invalidating the analysis window based on the point cloud quality field includes:

[0022] S5: Divide the target seam point cloud into multiple seam minimum analysis units along the main direction of the seam, and calculate the point cloud density, local missing rate, balance of points on both sides of the seam, normal dispersion and local fitting residual of each seam minimum analysis unit.

[0023] S6: Normalize and weight the point cloud density, local missing rate, point balance on both sides of the seam, normal dispersion, and local fitting residual to construct a point cloud quality field distributed along the main direction of the seam, and generate a quality-constrained seam analysis window based on the point cloud quality field; wherein, an initial window is generated with a preset initial window length; when the initial window does not meet the preset window quality conditions, the window is expanded along the main direction of the seam with the smallest analysis unit of the seam as the expansion step size; when the expanded window meets the preset window quality conditions, it is determined as a valid expanded window; when there are local low-quality units or local missing units between adjacent windows, and the merged window meets the preset window quality conditions, the adjacent windows are merged into a valid merged window; when the window reaches the preset maximum window length but still does not meet the preset window quality conditions, it is determined as an invalid window.

[0024] The calculation of misalignment and opening within the effective window, and the fusion of overlapping window results according to window confidence, includes:

[0025] S7: Within each effective quality constraint joint analysis window, separate the point clouds of the pipe segments on both sides of the joint according to the joint local reference coordinate system, and establish local surface models on both sides of the joint respectively.

[0026] S8: Under the local reference coordinate system of the joint, calculate the local misalignment based on the corresponding position difference of the local patch models on both sides of the joint on the misalignment direction axis, and calculate the local opening based on the projection distance of the feature points on both sides of the joint on the opening direction axis.

[0027] S9: Calculate the window confidence based on the point cloud integrity, point balance on both sides of the joint, local patch fitting residual, normal consistency, and deformation continuity of adjacent effective windows for each effective mass constraint joint analysis window.

[0028] S10: Remove windows with a confidence level lower than the preset confidence threshold, and perform confidence-weighted fusion of the local misalignment and local opening of multiple effective quality constraint joint analysis windows covering the same position in the main direction of the joint to obtain a fused misalignment sequence and a fused opening sequence distributed along the main direction of the joint.

[0029] Furthermore, the step of identifying continuous segments of misalignment anomalies, opening anomalies, and composite anomalies based on the fusion deformation includes:

[0030] S11: Based on the comparison results of the fused misalignment sequence and the fused opening sequence with the corresponding deformation threshold, the abnormal window is divided into misalignment abnormal window, opening abnormal window, or misalignment-opening composite abnormal window, and clustering is performed according to the spatial continuity condition and deformation continuity condition corresponding to different abnormal types; when there is a low confidence interval window between two abnormal candidate segments, and the length of the main direction of the joint corresponding to the low confidence interval window is less than the preset interval length threshold, and the abnormal candidate segments on both sides of the interval meet the corresponding deformation continuity condition, the two abnormal candidate segments are merged to obtain the joint deformation abnormal continuous segment;

[0031] S12: Output the starting position, ending position, section length, maximum misalignment, average misalignment, maximum opening, average opening, section confidence level, and anomaly type of the continuous section with abnormal joint deformation, as the detection result of the continuous section with abnormal joint deformation of the shield tunnel segment.

[0032] Extract the point cloud of the target joint and construct a local orthogonal reference coordinate system based on the main direction of the joint and the surface normal vectors of the segments on both sides of the joint;

[0033] The smallest analysis unit is divided along the main direction of the seam. A point cloud quality field is constructed based on the point cloud density, local missing rate, balance of points on both sides, normal dispersion and fitting residual. The analysis window is expanded, merged or invalidated based on the point cloud quality field.

[0034] Calculate the misalignment and opening within the effective window, and fuse the results of overlapping windows according to the window confidence level;

[0035] Identify continuous sections of misalignment anomalies, opening anomalies, and compound anomalies based on the fusion deformation.

[0036] Furthermore, the step of extracting the target joint point cloud and constructing a local orthogonal reference coordinate system based on the main direction of the joint and the surface normal vectors of the segments on both sides of the joint includes:

[0037] S1: Use laser scanning equipment to acquire three-dimensional point cloud data of the joint area of ​​shield tunnel segments;

[0038] S2: Perform outlier removal, voxel downsampling, normal vector estimation, and local smoothing on the three-dimensional point cloud data to obtain a standardized point cloud of the seam region;

[0039] S3: Extract candidate seam point clouds based on the local geometric abrupt change features of the standardized point cloud of the seam region, and perform connectivity screening on the candidate seam point clouds to obtain the target seam point cloud.

[0040] S4: Perform robust principal direction analysis on the target joint point cloud to obtain the joint principal direction axis; perform direction consistency correction and weighting on the average normal vectors of the segments on both sides of the joint, and orthogonalize the weighted direction vectors relative to the joint principal direction to obtain the misalignment direction axis; obtain the opening direction axis based on the cross product direction of the joint principal direction axis and the misalignment direction axis, thereby constructing a mutually orthogonal local reference coordinate system for the joint.

[0041] Furthermore, the step of dividing the analysis window into minimum analysis units along the main direction of the seam, constructing a point cloud quality field based on point cloud density, local missing rate, balance of points on both sides, normal dispersion, and fitting residual, and then expanding, merging, or invalidating the analysis window based on the point cloud quality field includes:

[0042] S5: Divide the target seam point cloud into multiple seam minimum analysis units along the main direction of the seam, and calculate the point cloud density, local missing rate, balance of points on both sides of the seam, normal dispersion and local fitting residual of each seam minimum analysis unit.

[0043] S6: Normalize and weight the point cloud density, local missing rate, point balance on both sides of the seam, normal dispersion, and local fitting residual to construct a point cloud quality field distributed along the main direction of the seam, and generate a quality-constrained seam analysis window based on the point cloud quality field; wherein, an initial window is generated with a preset initial window length; when the initial window does not meet the preset window quality conditions, the window is expanded along the main direction of the seam with the smallest analysis unit of the seam as the expansion step size; when the expanded window meets the preset window quality conditions, it is determined as a valid expanded window; when there are local low-quality units or local missing units between adjacent windows, and the merged window meets the preset window quality conditions, the adjacent windows are merged into a valid merged window; when the window reaches the preset maximum window length but still does not meet the preset window quality conditions, it is determined as an invalid window.

[0044] Furthermore, the step of calculating the misalignment and opening within the effective window, and fusing the overlapping window results according to the window confidence level, includes:

[0045] S7: Within each effective quality constraint joint analysis window, separate the point clouds of the pipe segments on both sides of the joint according to the joint local reference coordinate system, and establish local surface models on both sides of the joint respectively.

[0046] S8: Under the local reference coordinate system of the joint, calculate the local misalignment based on the corresponding position difference of the local patch models on both sides of the joint on the misalignment direction axis, and calculate the local opening based on the projection distance of the feature points on both sides of the joint on the opening direction axis.

[0047] S9: Calculate the window confidence based on the point cloud integrity, point balance on both sides of the joint, local patch fitting residual, normal consistency, and deformation continuity of adjacent effective windows for each effective mass constraint joint analysis window.

[0048] S10: Remove windows with a confidence level lower than the preset confidence threshold, and perform confidence-weighted fusion of the local misalignment and local opening of multiple effective quality constraint joint analysis windows covering the same position in the main direction of the joint to obtain a fused misalignment sequence and a fused opening sequence distributed along the main direction of the joint.

[0049] Furthermore, the step of identifying continuous segments of misalignment anomalies, opening anomalies, and composite anomalies based on the fusion deformation includes:

[0050] S11: Based on the comparison results of the fused misalignment sequence and the fused opening sequence with the corresponding deformation threshold, the abnormal window is divided into misalignment abnormal window, opening abnormal window, or misalignment-opening composite abnormal window, and clustering is performed according to the spatial continuity condition and deformation continuity condition corresponding to different abnormal types; when there is a low confidence interval window between two abnormal candidate segments, and the length of the main direction of the joint corresponding to the low confidence interval window is less than the preset interval length threshold, and the abnormal candidate segments on both sides of the interval meet the corresponding deformation continuity condition, the two abnormal candidate segments are merged to obtain the joint deformation abnormal continuous segment;

[0051] S12: Output the starting position, ending position, section length, maximum misalignment, average misalignment, maximum opening, average opening, section confidence level, and anomaly type of the continuous section with abnormal joint deformation, as the detection result of the continuous section with abnormal joint deformation of the shield tunnel segment.

[0052] Furthermore, in S4, the joint local reference coordinate system includes the joint principal direction axis, the misalignment direction axis, and the opening direction axis;

[0053] The average normal vectors of the pipe segments on both sides of the joint are corrected for directional consistency so that they point to the same side.

[0054] Based on the corrected normal vectors on both sides, the initial misalignment direction vector is calculated by weighting with the corresponding mass weights. Then, the initial misalignment direction vector is orthogonalized relative to the main direction axis of the joint to obtain the misalignment direction axis.

[0055] The opening direction axis is determined by the cross product of the main direction axis of the joint and the misalignment direction axis;

[0056] The coordinates of any point in the target seam point cloud under the local reference coordinate system of the seam are determined by the spatial coordinates of the point, the origin of the coordinate system, and the direction vectors of each coordinate axis.

[0057] Furthermore, in S5, the point cloud density is determined by the number of effective points within the smallest analysis unit of the seam, the unit length, and the bandwidth participating in the analysis on both sides of the seam.

[0058] The local missing rate is determined by the ratio of the length of the seam direction covered by the effective point cloud within the smallest analysis unit of the seam to the length of the unit.

[0059] The balance of the number of points on both sides of the joint is determined by the degree of difference in the number of effective points on the left and right sides of the joint within the unit; the smaller the difference, the higher the balance.

[0060] Furthermore, in S5, the normal dispersion is determined by the statistical value of the deviation between the local normal vector of each point in the smallest analysis unit of the joint and the average normal vector of the unit;

[0061] The local fitting residual is determined by the statistical values ​​of the distances from each point within the smallest analysis unit of the seam to the local patch fitting model.

[0062] Furthermore, in S6, the point cloud quality value of each unit in the point cloud quality field is obtained by weighted summation of the normalized point cloud density, local missing rate, point balance, normal dispersion, and local fitting residual.

[0063] The normalization of each parameter adopts a threshold truncation method, and is normalized with preset reference values ​​respectively;

[0064] When the number of valid points in a certain cell is lower than the preset minimum number of points, the point cloud quality value of that cell is set to zero and marked as a missing cell.

[0065] Furthermore, in S6, generating a mass constraint seam analysis window based on the point cloud mass field includes:

[0066] Multiple initial windows are generated sequentially along the main direction of the seam with a preset initial window length and a sliding step size smaller than that length, with adjacent windows partially overlapping;

[0067] The overall quality value of a window is determined by the ratio of the sum of the point cloud quality values ​​of the smallest analysis units of each seam contained therein to the number of units.

[0068] An effective window must simultaneously meet the following requirements: the overall quality value is not lower than the preset quality threshold, the number of effective points is not lower than the preset minimum number of points, the local missing rate is not higher than the preset maximum allowable missing rate, and the balance of the number of points on both sides of the seam is not lower than the preset minimum allowable balance.

[0069] If the initial window does not meet the conditions, it is gradually expanded with the smallest analysis unit of the seam as the step size until the conditions are met or the preset maximum window length is reached; the expanded window that meets the conditions is considered a valid window.

[0070] When there are local low-quality or missing units between adjacent windows, but the merged window can meet the valid window conditions, the two are merged into one valid window;

[0071] If the window reaches its maximum length but still does not meet the conditions, or if the missing rate within the window continues to exceed the limit, it is determined to be an invalid window and will not participate in subsequent deformation calculations.

[0072] Furthermore, the local misalignment amount is determined by the coordinate difference between the local surface models on the left and right sides of the joint at the center position along the misalignment direction axis.

[0073] The local opening amount is determined by the difference between the average projected coordinates of the feature points on the left and right edges of the joint along the opening direction axis.

[0074] Furthermore, in S9, the window confidence is obtained by weighted summation of point cloud integrity confidence, seam side balance confidence, patch fitting confidence, normal consistency confidence, and adjacent window continuity confidence.

[0075] Among them, the confidence level of point cloud integrity is determined based on the local missing rate of the window, the confidence level of the balance on both sides of the seam is determined based on the point number balance, the confidence level of patch fitting is determined based on the local patch fitting residual, the confidence level of normal consistency is determined based on the normal dispersion, and the confidence level of the continuity of adjacent windows is determined based on the exponential decay function of the difference in misalignment and the difference in opening between the current window and the previous valid window; when there is no preceding valid window, the continuity confidence level takes a preset value;

[0076] For invalid windows that have been identified, their window confidence is set to zero and they are not included in subsequent calculations.

[0077] Furthermore, in S10, the fusion misalignment amount and fusion opening amount are obtained by weighted averaging of the local misalignment amount or local opening amount of all effective quality constraint joint analysis windows at the same position in the main direction of the covered joint, with the confidence level of each window as the weight.

[0078] Furthermore, in S11, the spatial continuity condition for adjacent abnormal windows means that the distance between adjacent abnormal windows along the main direction of the seam is less than a preset spatial continuity threshold.

[0079] For adjacent windows that are both misaligned, the continuity condition for deformation is that the absolute value of the difference between their merged misalignment is less than the misalignment mutation threshold. For adjacent windows that are both opening anomalies, the condition is that the absolute value of the difference between their merged opening is less than the opening mutation threshold. For adjacent windows that are both compound anomalies, the continuity conditions for both misalignment and opening must be met simultaneously.

[0080] When adjacent windows have different anomaly types and one of them is a compound anomaly, if the adjacent windows meet the corresponding continuity conditions on the deformation index that both exceed the threshold, they can be classified into the same segment; if one is a misalignment anomaly and the other is an opening anomaly, they will not be classified into the same segment.

[0081] When two abnormal candidate segments are separated by a low-confidence interval window, and the length of the seam direction of the interval window is less than the preset interval length threshold, and the adjacent ends of the two segments meet the corresponding deformation continuity conditions, the two segments are merged.

[0082] Furthermore, in S12, the output of the joint deformation anomaly continuous section detection results includes the starting position, ending position, section length, maximum misalignment, average misalignment, maximum opening, average opening, section confidence level, and anomaly type of the section.

[0083] The confidence level of a segment is obtained by averaging the confidence levels of all valid quality constraint seam analysis windows participating in the statistics within that segment; when a segment is formed by merging candidate segments on both sides of a low-confidence interval window, that interval window is not included in the statistics.

[0084] Furthermore, the method for determining the anomaly type is as follows: when the maximum misalignment amount within a segment exceeds a preset misalignment threshold and the maximum opening amount does not exceed a preset opening threshold, it is determined to be a misalignment anomaly; when the maximum opening amount exceeds a preset opening threshold and the maximum misalignment amount does not exceed a misalignment threshold, it is determined to be an opening anomaly; when the maximum misalignment amount exceeds a misalignment threshold and the maximum opening amount exceeds an opening threshold, it is determined to be a misalignment-opening composite anomaly.

[0085] The advantages of this invention are:

[0086] First, this invention addresses the problems of skewed scanning posture, uneven local density, and inconsistent data quality on both sides of the joint that are prone to occur when handheld laser scanning point clouds. It constructs a local orthogonal reference coordinate system for the joint based on the main direction of the joint and the surface normal vectors of the segments on both sides of the joint after direction consistency correction. This eliminates the need to rely on the complete tunnel cross-section, the global tunnel centerline, or the center of the fitted cross-section, which helps to improve the stability of the measurement benchmark for local joint deformation.

[0087] Second, this invention constructs a point cloud quality field distributed along the main direction of the joint based on point cloud density, local missing rate, balance of points on both sides of the joint, normal dispersion, and local fitting residual, so that the reliability of data at different joint locations can be quantitatively characterized, providing a data quality basis for subsequent analysis window generation and measurement result screening.

[0088] Third, this invention generates a quality-constrained seam analysis window based on the point cloud quality field. When the initial window does not meet the preset window quality conditions, it is expanded with the smallest seam analysis unit as the step size; when there are local low-quality units or local missing units between adjacent windows and the window quality conditions are met after merging, the adjacent windows are merged; when the window quality conditions are still not met after reaching the preset maximum window length, the corresponding window is determined to be an invalid window. Therefore, the risk of unreliable measurement results generated by a fixed-length window in areas of varying point cloud density and areas of local missing units can be reduced.

[0089] Fourth, within the same joint local orthogonal reference coordinate system and the same mass-constrained joint analysis window, the present invention calculates the local misalignment based on the local patch models on both sides of the joint, and calculates the local opening based on the edge feature points on both sides of the joint. This allows for the simultaneous acquisition of joint misalignment deformation and opening deformation, avoiding the need to evaluate the segment joint status using only a single misalignment index.

[0090] Fifth, this invention calculates the window confidence based on point cloud integrity, point balance on both sides of the seam, local patch fitting residual, normal consistency, and continuity of deformation of adjacent effective windows, and performs confidence-weighted fusion of measurement results of multiple effective windows at the same position in the main direction of the seam, which helps to reduce deformation fluctuations caused by local occlusion, missing point clouds, outliers, and local fitting errors.

[0091] Sixth, this invention divides the abnormal window into misaligned abnormal window, open abnormal window, and misaligned-open composite abnormal window based on the fusion misalignment amount and fusion opening amount. It also performs clustering based on the corresponding spatial continuity and deformation continuity conditions according to the abnormality type. For abnormal candidate segments separated by a small number of low-confidence interval windows and with continuous deformation trends on both sides of the interval, it can perform condition merging, thereby reducing the missegmentation of continuous defects caused by local point cloud missingness, and outputs the start and end positions, length, deformation statistics, segment confidence and abnormality type of the abnormal segment. Attached Figure Description

[0092] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0093] Figure 1 This is a flowchart illustrating the overall process of the method of the present invention.

[0094] Figure 2 A schematic diagram of point cloud acquisition of the joint area of ​​shield tunnel segments using handheld laser scanning.

[0095] Figure 3A schematic diagram of constructing a local reference coordinate system for the joint.

[0096] Figure 4 A schematic diagram is generated for the smallest analysis unit of the joint, the point cloud mass field, and the mass constraint joint analysis window.

[0097] Figure 5 This is a schematic diagram showing the local surface model on both sides of the joint and the calculation of local misalignment and local opening.

[0098] Figure 6 A schematic diagram for calculating window confidence and generating deformation sequence.

[0099] Figure 7 A schematic diagram showing the formation of continuous segments with abnormal joint deformation through clustering of abnormal windows and the output results. Detailed Implementation

[0100] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0101] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. It should be understood that the following embodiments are only for illustrating the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention. Without departing from the concept of the present invention, those skilled in the art can make equivalent substitutions or appropriate adjustments to the relevant steps, parameters, thresholds, and algorithm implementations.

[0102] like Figure 1 As shown in the figure, an embodiment of the present invention provides a method for detecting continuous deformation sections of shield tunnel segment joints, including steps S1 to S12.

[0103] In step S1, the inspector moves a handheld laser scanning device along the joint area of ​​the tunnel segments to be inspected, acquiring three-dimensional point cloud data within a certain width range near the joint. The three-dimensional point cloud data includes the three-dimensional coordinates of the points, and may also include information such as acquisition time, scanning sequence, point intensity, or device posture. Because the inspector's movement speed, device angle, and scanning distance may change during the handheld scanning process, the acquired point cloud along the joint direction typically exhibits uneven density and localized gaps.

[0104] In S2, the original 3D point cloud is preprocessed. Preprocessing includes outlier removal, voxel downsampling, normal vector estimation, and local smoothing. Outlier removal removes isolated noise points far from the segment surface; voxel downsampling reduces redundant points while preserving the main geometric features of the joint area; normal vector estimation obtains the local orientation of the segment surface; and local smoothing reduces the impact of handheld scanning noise on subsequent geometric change identification.

[0105] In S3, candidate joint point clouds are extracted based on the local geometric abrupt change features of the standardized point cloud in the joint area. Since segment joints typically correspond to geometrically discontinuous regions on the surface, the point cloud near the joint exhibits abrupt changes in curvature, normal vector variation, or local point distribution. Candidate joint point clouds are obtained by screening regions with high local curvature, significant normal vector changes, or depressions in the neighborhood point cloud distribution. Subsequently, connectivity screening is performed on the candidate joint point clouds to remove scattered noise points and non-joint abrupt change points, yielding the target joint point cloud.

[0106] In step S4, robust principal direction analysis is performed on the target seam point cloud to obtain the seam principal direction. Robust principal direction analysis can be achieved using principal component analysis, random sampling consensus linear fitting, or weighted direction estimation methods. Considering the potential presence of noise points and local missing data in the target seam point cloud, this embodiment preferably uses weighted principal direction analysis to reduce the impact of outliers on the estimation of the seam principal direction.

[0107] In S5, the target seam point cloud is divided into multiple seam minimum analysis units along the main direction of the seam. Preferably, the length of the seam minimum analysis unit is... The bandwidth of the joint on both sides is 50mm to 200mm. The range is 100mm to 300mm. For each smallest analysis unit of the seam, the point cloud density, local missing rate, balance of points on both sides of the seam, normal dispersion, and local fitting residual are calculated respectively.

[0108] In S6, a point cloud quality field is constructed based on the point cloud quality parameters, and a quality constraint seam analysis window is generated based on the point cloud quality field. Preferably, the initial window length is 100 mm to 300 mm, the maximum window length is 300 mm to 800 mm, and the minimum number of effective points is... The maximum allowable missing value is 30-100. The minimum allowable balance is set at 0.4 to 0.6. The value is set to 0.5 to 0.7. When the initial window meets the valid window condition, it is used as the valid quality constraint joint analysis window; when the initial window does not meet the valid window condition, the window is expanded along the main direction of the joint with the smallest analysis unit of the joint as the expansion step size until the valid window condition is met or the preset maximum window length is reached; when the expanded window meets the valid window condition, it is used as the expanded window; when there are local low-quality units or local missing units between adjacent windows and the merged window meets the valid window condition, the adjacent windows are merged to form a merged window; when the window still cannot meet the valid window condition after reaching the preset maximum window length, it is determined as an invalid window and is not included in the subsequent calculation of local misalignment and local opening.

[0109] In S7, within the effective window, the point cloud is divided into point clouds of segments on both sides of the joint based on the local reference coordinate system of the joint. The specific division can be based on the sign of the point cloud on the opening direction axis, the position of the joint centerline, or the position of edge feature points. For the point clouds on both sides of the joint, local patch models are established separately. These local patch models can be local planar models, quadratic surface models, or locally robust patch models.

[0110] In the process of establishing the local patch model, point cloud quality parameters are used as weights in the fitting process. Points with higher point cloud density, better normal consistency, and smaller local residuals are assigned higher weights; outliers, occluded edge points, and points with larger residuals are assigned lower weights or are removed. This results in a relatively stable local patch model on both sides of the seam.

[0111] In S8, the local misalignment and local opening of each effective window are calculated in the local reference coordinate system of the joint. The local misalignment is determined by the coordinate difference along the misalignment direction axis at the corresponding positions of the local patch models on both sides of the joint near the joint edge; the local opening is determined by the projection distance of the feature points on both sides of the joint edge along the opening direction axis.

[0112] In S9, a window confidence score is calculated for each valid window. The window confidence score is jointly determined by the point cloud integrity confidence score, the seam balance confidence score, the patch fitting confidence score, the normal consistency confidence score, and the adjacent window continuity confidence score. Preferably, a window confidence score threshold is used. Take a value of 0.5 to 0.8.

[0113] In S10, when the confidence level of a window is lower than a preset confidence threshold, the window is marked as an invalid window and does not participate in the generation of the final deformation sequence. When the confidence level of a window meets the requirements, its local misalignment and local opening are fused according to the confidence level to obtain a deformation sequence distributed along the main direction of the joint.

[0114] In S11, abnormal windows are identified based on preset misalignment thresholds and opening thresholds. If the fusion misalignment amount at the corresponding position of a window exceeds the misalignment threshold but the fusion opening amount does not exceed the opening threshold, the window is marked as a misalignment abnormal window; if the fusion opening amount at the corresponding position of a window exceeds the opening threshold but the fusion misalignment amount does not exceed the misalignment threshold, the window is marked as an opening abnormal window; if both the fusion misalignment amount and the fusion opening amount exceed the corresponding thresholds, the window is marked as a misalignment-opening composite abnormal window.

[0115] When clustering abnormal windows, spatial continuity, anomaly type consistency, and deformation continuity are considered simultaneously. For adjacent abnormal windows, first, it is determined whether their spatial distance along the main direction of the joint is less than a preset spatial continuity threshold. Based on the spatial continuity condition, the deformation continuity is determined according to the abnormal window type. For misaligned abnormal windows, it is determined whether the difference in fused misalignment between adjacent windows is less than the misalignment mutation threshold. For open abnormal windows, it is determined whether the difference in fused opening between adjacent windows is less than the opening mutation threshold. For misaligned-open composite abnormal windows, it is determined whether both the difference in fused misalignment and the difference in fused opening between adjacent windows meet the corresponding mutation threshold requirements. If two candidate abnormal segments are separated only by a small number of low-confidence windows, and the interval length is less than a preset interval length threshold, and the deformation trends of the candidate abnormal segments on both sides of the interval meet the corresponding deformation continuity condition, then the two candidate abnormal segments are merged to avoid missegmentation of continuous defects due to missing local point clouds.

[0116] In S12, the start and end positions, segment length, maximum misalignment, average misalignment, maximum opening, average opening, segment confidence level, and anomaly type of each abnormal continuous segment are output. The start and end positions can be represented by joint local coordinates, tunnel mileage, segment ring number, or distance relative to the scanning start point. The segment confidence level can be determined by the average or weighted average of the confidence levels of each valid window within the segment.

[0117] Through the above steps, this embodiment can stably obtain continuous sections with abnormal deformation along the joints of shield tunnel segments even when there are uneven density, local missing parts, and posture deviation in the handheld scanning point cloud. It is suitable for rapid inspection, key defect verification, and defect development trend comparison in shield tunnel operation and maintenance.

[0118] The following provides a detailed explanation of how each step is implemented:

[0119] S1: Use a handheld laser scanning device to acquire three-dimensional point cloud data of the joint area of ​​the shield tunnel segments. The three-dimensional point cloud data includes the spatial coordinates of the points and the point cloud acquisition sequence information corresponding to the scanning process.

[0120] S2: Perform outlier removal, voxel downsampling, normal vector estimation, and local smoothing on the three-dimensional point cloud data to obtain a standardized point cloud of the seam region.

[0121] S3: Extract candidate seam point clouds based on the local geometric abrupt change features of the standardized point cloud of the seam region, and perform connectivity screening on the candidate seam point clouds to obtain the target seam point cloud.

[0122] S4: Perform robust principal direction analysis based on the target joint point cloud to obtain the joint principal direction, and construct a local reference coordinate system for the joint based on the joint principal direction, the normal direction of the pipe segments on both sides of the joint, and the lateral opening direction of the joint.

[0123] More preferably, in S4, the joint local reference coordinate system includes a joint principal direction axis, a misalignment direction axis, and a spreading direction axis. The unit direction vector of the joint principal direction axis is denoted as... To avoid instability in the calculation of the misalignment direction axis due to inconsistent normal vector directions on both sides of the joint, the average normal vector of the left side of the joint segment surface is calculated before constructing the local reference coordinate system of the joint. and the average normal vector of the right side of the segment surface of the joint Perform orientation consistency correction to make them point to the same side; when and When the included angle is greater than 90°, one of the average normal vectors is inverted.

[0124] Based on the average normal vectors of the left and right segments of the joint after orientation consistency correction, the initial misalignment direction vector is calculated according to the following formula. :

[0125]

[0126] in, Assign quality weights to the point cloud on the left side of the seam. The quality weight of the point cloud on the right side of the seam.

[0127] For the initial misalignment direction vector Orthogonalization is performed relative to the main direction axis of the joint to obtain the unit direction vector of the misalignment direction axis. The unit direction vector of the opening direction axis is determined based on the main direction axis of the joint and the misalignment direction axis. They are represented by the following formulas:

[0128]

[0129]

[0130] The coordinates of any point in the target seam point cloud in the local reference coordinate system of the seam are expressed by the following formula:

[0131]

[0132] In the formula: The unit direction vector of the main direction axis of the joint; Let this be the initial misalignment direction vector; The unit direction vector of the misalignment direction axis; The unit direction vector of the opening direction axis; The average normal vector of the segment surface on the left side of the joint; The average normal vector of the segment surface on the right side of the joint; For the target seam point cloud, the first The spatial coordinates of the points; The origin of the local reference coordinate system for the joint; For point Coordinates along the main direction axis of the joint; For point Coordinates along the misalignment axis; For point Coordinates along the opening direction axis.

[0133] S5: Divide the target seam point cloud into multiple seam minimum analysis units along the main direction of the seam, and calculate the point cloud quality parameters of each seam minimum analysis unit. The point cloud quality parameters include point cloud density, local missing rate, balance of point count on both sides of the seam, normal dispersion, and local fitting residual.

[0134] More preferably, in S5, the point cloud density, local missing rate, and point count balance on both sides of the seam are expressed by the following formulas:

[0135]

[0136]

[0137]

[0138] In the formula: For the first Point cloud density of the smallest analysis unit of a seam; For the first The number of valid points within the smallest analysis unit of a seam; The minimum analysis unit length for the joint; The bandwidth involved in the analysis on both sides of the joint; For the first Local missing rate of the smallest analysis unit of each seam; For the first The length of the seam direction within the smallest analysis unit of the seam has effective point cloud coverage; For the first Balance of the number of points on both sides of the joint in the smallest analysis unit of a joint; For the first Number of valid points on the left side of the joint within the smallest analysis unit of each joint; For the first Number of valid points on the right side of the joint within the smallest analysis unit of each joint; To prevent extremely small positive numbers with a denominator of zero.

[0139] More preferably, in S5, the normal dispersion and the local fitting residual are expressed by the following formulas:

[0140]

[0141]

[0142] In the formula: For the first Normal dispersion of the smallest analytical unit of a seam; For the first The smallest analysis unit for each seam; For point The local normal vector; For the first The average normal vector of the point cloud within the smallest analysis unit of each seam; For the first Local fitting residuals of the smallest analysis unit of each seam; For the first Local patch fitting model within the smallest analysis unit of a seam; For point To local patch fitting model The distance.

[0143] S6: Construct a point cloud quality field distributed along the main direction of the joint based on the point cloud quality parameters, and generate a quality constraint joint analysis window based on the point cloud quality field.

[0144] In S6, the point cloud mass field of the first... The point cloud quality value of the smallest analysis unit for each seam is expressed by the following formula:

[0145]

[0146] in:

[0147]

[0148] and:

[0149]

[0150] To avoid point cloud quality parameters with different dimensions directly participating in the weighted calculation, point cloud density, local missing rate, point number balance on both sides of the seam, normalization is performed on the point cloud density, local missing rate, normalization dispersion, and local fitting residual before constructing the point cloud quality field. The normalization process uses a threshold-truncated normalization method, expressed by the following formulas:

[0151]

[0152]

[0153]

[0154]

[0155]

[0156] When the Number of valid points within the smallest analysis unit of a seam Less than the preset minimum number of valid points At that time, the point cloud quality value of the smallest analysis unit of the joint is... Set it to 0 and mark the smallest analysis unit of the joint as a missing unit.

[0157] In the formula: For the first Point cloud quality value of the smallest analysis unit of a seam; For the first Point cloud density of the smallest analysis unit of a seam; For the first Local missing rate of the smallest analysis unit of each seam; For the first Balance of the number of points on both sides of the joint in the smallest analysis unit of a joint; For the first Normal dispersion of the smallest analytical unit of a seam; For the first Local fitting residuals of the smallest analysis unit of each seam; This represents the normalized point cloud density. This represents the normalized local deletion rate; The balance of the number of points on both sides of the joint after normalization; This represents the normalized normal dispersion. The normalized local fitting residuals; This serves as a reference value for the normalization of point cloud density. This serves as a reference value for the normalization of local missing data rates. This serves as a reference value for the normalization of the normal dispersion. This serves as a reference value for the normalization of local fitting residuals. The minimum number of valid points in the preset unit; , , , , These are the weighting coefficients for the corresponding point cloud quality parameters.

[0158] A quality-constrained joint analysis window is considered a valid window when it satisfies the following formula:

[0159]

[0160]

[0161]

[0162]

[0163] In the formula: For the first A quality constraint joint analysis window; For the first The overall quality value of each quality-constrained joint analysis window; For the first The number of minimum analysis units for a joint contained within a mass constraint joint analysis window; For the first Point cloud quality value of the smallest analysis unit of a seam; For the first The smallest analysis unit for each seam; The preset window quality threshold; For the first The number of valid points within a quality constraint joint analysis window. The minimum number of valid points is preset; For the first Local missing rate of each quality constraint seam analysis window; The maximum allowed missing rate is preset; For the first Balance of the number of points on both sides of the joint in the quality constraint joint analysis window; This is the preset minimum allowable balance.

[0164] S7: Within each valid quality constraint joint analysis window, separate the point clouds of the pipe segments on both sides of the joint according to the joint local reference coordinate system, and establish local surface models on both sides of the joint respectively.

[0165] S8: Under the local reference coordinate system of the joint, calculate the local misalignment and local opening of each effective mass constraint joint analysis window.

[0166] More preferably, in S8, the local misalignment and local opening are expressed by the following formulas:

[0167]

[0168]

[0169] In the formula: For the first Local misalignment within an effective quality constraint joint analysis window; For the first Local opening of an effective quality constraint joint analysis window; This is a partial surface model of the left side of the seam; This is a partial surface model of the right side of the seam; For the first The coordinates of the center position of the effective quality constraint joint analysis window on the main direction axis of the joint; The coordinates of the corresponding position near the seam edge on the opening direction axis; For the first The average projected coordinates of the feature points on the left edge of the joint within the effective quality constraint joint analysis window on the opening direction axis; For the first The average projected coordinates of the right edge feature points of the joint on the opening direction axis within the effective quality constraint joint analysis window.

[0170] S9: Calculate the window confidence based on the point cloud integrity, point balance on both sides of the joint, local patch fitting residual, normal consistency, and deformation continuity of adjacent effective windows within each quality constraint joint analysis window.

[0171] In S9, the window confidence level is expressed by the following formula:

[0172]

[0173] in:

[0174]

[0175]

[0176]

[0177]

[0178]

[0179]

[0180] In the formula: For the first Window confidence of a quality constraint joint analysis window; Confidence level for point cloud integrity; To balance the confidence levels on both sides of the joint; Confidence level for patch fitting; For normal consistency confidence; The confidence level for continuity between adjacent windows; These are the weighting coefficients for the corresponding confidence components; For the first Local missing rate of each quality constraint seam analysis window; For the first Balance of the number of points on both sides of the joint in the quality constraint joint analysis window; For the first Local patch fitting residuals within a quality-constrained joint analysis window; For the first Normal dispersion within a mass-constrained joint analysis window; Indicates the direction along the main direction of the joint and the first The sequence number of the previous effective mass constraint joint analysis window adjacent to the current effective mass constraint joint analysis window. and The first The local misalignment of each effective mass constraint joint analysis window and its preceding effective mass constraint joint analysis window. and The first The local opening of each effective mass constraint joint analysis window and its preceding effective mass constraint joint analysis window. These are the adjustment coefficients for the corresponding exponential functions.

[0181] When the When a valid mass constraint joint analysis window does not exist in the previous valid mass constraint joint analysis window, let For the quality constraint joint analysis window that was determined to be invalid in S6, set its window confidence level. Furthermore, it does not participate in subsequent calculations of local misalignment, local opening, or confidence-weighted fusion.

[0182] S10: Remove invalid windows with a confidence level lower than the preset confidence threshold, and perform confidence-weighted fusion of the local misalignment and local opening of the remaining valid windows to obtain a deformation sequence distributed along the main direction of the target joint.

[0183] More preferably, in S10, the fusion misalignment and fusion opening in the deformation sequence are represented by the following formulas:

[0184]

[0185]

[0186] In the formula: Position of the main direction of the seam The amount of misalignment at the fusion point; Position of the main direction of the seam The fusion opening amount at the location; To cover the main direction of the joint A set of effective quality constraint joint analysis windows; For the first Window confidence of an effective quality constraint joint analysis window; For the first Local misalignment within an effective quality constraint joint analysis window; For the first Local opening of an effective quality constraint joint analysis window; To prevent extremely small positive numbers with a denominator of zero.

[0187] S11: Mark windows in the deformation sequence that exceed the corresponding deformation threshold as abnormal windows, and cluster them according to the spatial continuity of the abnormal windows in the main direction of the joint, the consistency of the abnormal type, and the continuity of the deformation amount to obtain the joint deformation abnormal continuous segment.

[0188] Specifically, the abnormal windows include misaligned abnormal windows, open abnormal windows, and a combination of misaligned and open abnormal windows. Adjacent abnormal windows satisfy the following spatial continuity condition:

[0189]

[0190] Further, determine the continuity of deformation based on the anomaly type of adjacent abnormal windows. If both adjacent abnormal windows are misalignment abnormal windows, then determine whether they meet the following misalignment continuity condition:

[0191]

[0192] If both adjacent abnormal windows are open abnormal windows, then determine whether they satisfy the following openness continuity condition:

[0193]

[0194] If adjacent abnormal windows are both misalignment-opening composite abnormal windows, then it is determined whether they simultaneously satisfy the continuity conditions of misalignment and opening. When both the spatial continuity condition and the corresponding deformation continuity condition of adjacent abnormal windows are satisfied, the adjacent abnormal windows are divided into the same joint deformation abnormality continuous section.

[0195] When a low-confidence interval window exists between two candidate abnormal segments, and the main direction length of the joint corresponding to the low-confidence interval window is less than a preset interval length threshold, and the abnormality type at the adjacent ends of the two candidate abnormal segments satisfies the corresponding deformation continuity condition, the two candidate abnormal segments are merged into the same joint deformation abnormality continuous segment. Where: For the first The first abnormal window and the first Spatial distance between abnormal windows along the main direction of the seam; The threshold for spatial continuity; and The first The first abnormal window and the first The amount of fusion misalignment at the location corresponding to each abnormal window; and The first The first abnormal window and the first The fusion opening amount at the corresponding position of each abnormal window; The threshold for sudden changes in misalignment; This is the threshold for the sudden change in opening amount.

[0196] S12: Output the starting position, ending position, section length, maximum misalignment, average misalignment, maximum opening, average opening, section confidence level, and anomaly type of the continuous section with abnormal joint deformation, as the detection result of the continuous section with abnormal joint deformation of the shield tunnel segment.

[0197] In S12, the detection result of the abnormal deformation continuous section of the joint is expressed by the following formula:

[0198]

[0199] in:

[0200]

[0201]

[0202]

[0203]

[0204]

[0205]

[0206] In the formula: For the first The detection results of a continuous section with abnormal joint deformation; For the first The starting point of a continuous section with abnormal joint deformation; For the first The endpoint of a continuous section with abnormal joint deformation; For the first The length of a continuous section with abnormal joint deformation; For the first The maximum misalignment in a continuous section of joint deformation abnormality. For the first The average misalignment of a continuous section with abnormal joint deformation; For the first The maximum opening of a continuous section with abnormal joint deformation; For the first The average opening of a continuous section of joint with abnormal deformation; For the first Segment confidence of continuous sections with abnormal joint deformation; For the first The anomaly type of a continuous section with abnormal joint deformation; For the first The set of main direction locations of joints involved in deformation statistics within a continuous section of abnormal joint deformation; The number of positions in the set of main direction positions of the joint; For the first The set of effective quality constraint joint analysis windows for participating in the section confidence statistics within a continuous section of abnormal joint deformation; The number of windows in the set of effective quality constraint joint analysis windows; For the first The window confidence level of each effective quality constraint joint analysis window. When a continuous segment of joint deformation anomaly is formed by merging anomaly candidate segments on both sides of a low-confidence interval window, the low-confidence interval window is not included. .

[0207] Figure 1 This is a flowchart illustrating the overall process of the method of this invention. Figure 1 As shown, the method of the present invention includes, in sequence: acquiring handheld laser scanning point cloud, point cloud preprocessing, extraction of target seam point cloud, construction of seam local reference coordinate system, division of seam minimum analysis unit and calculation of quality parameters, construction of point cloud mass field and generation of mass constraint seam analysis window, separation of point cloud on both sides of seam and local patch modeling, calculation of local misalignment and local opening, calculation of window confidence, deformation sequence fusion, identification of abnormal continuous sections, and output of detection results.

[0208] Figure 2This is a schematic diagram of point cloud acquisition of shield tunnel segment joint area using handheld laser scanning. The diagram shows the handheld laser scanning equipment, shield tunnel segment, target joint area, and non-uniform point cloud formed by changes in scanning posture.

[0209] Figure 3 A schematic diagram of the construction of a local reference coordinate system for the joint is shown in the figure, which shows the target joint point cloud, the main direction axis of the joint, the misalignment direction axis, and the opening direction axis.

[0210] Figure 4 A schematic diagram is generated for the joint minimum analysis unit, point cloud mass field, and mass-constrained joint analysis window. For example... Figure 4 As shown, the target seam point cloud is divided into multiple seam minimum analysis units 7 along the main direction u of the seam. A point cloud quality field 8 is constructed based on the point cloud density, local missing rate, point balance on both sides of the seam, normal dispersion, and local fitting residual of each seam minimum analysis unit 7. The point cloud quality field 8 is used to characterize the reliability of the point cloud data at different locations along the seam, where different filling methods represent high quality, medium quality, low quality, and very low quality or missing states, respectively. A quality constraint seam analysis window 9 is generated based on the point cloud quality field 8. The quality constraint seam analysis window 9 includes an initial window 91, an expanded window 92, a merged window 93, and an invalid window 94. When the point cloud quality in the initial window 91 meets the preset conditions, it is used as a valid analysis window. When the number of valid points, the local missing rate, or the balance of the number of points on both sides of the seam in the initial window 91 does not meet the preset conditions, it is expanded along the main direction u of the seam to form an expanded window 92. When there are local low quality or local missing points between adjacent windows but the overall quality constraint conditions are still met, a merged window 93 is formed. When the point cloud quality in the window is continuously lower than the preset quality threshold or there are serious local missing points, it is determined to be an invalid window 94, and the invalid window 94 does not participate in the subsequent calculation of local misalignment and local opening.

[0211] Figure 5 The diagram shows the local surface model and deformation calculation of the joint on both sides. The diagram shows the point cloud of the pipe segments on both sides of the joint, the local surface model on both sides, the local misalignment, and the local opening.

[0212] Figure 6 The diagram illustrates the calculation of window confidence and the generation of deformation sequence. The diagram shows multiple valid windows, the confidence of each window, the local misalignment, the local opening, and the deformation sequence after confidence fusion.

[0213] Figure 7 The diagram illustrates the formation of continuous segments of abnormal joint deformation through clustering of abnormal windows and the output results. The diagram shows abnormal windows exceeding the threshold, low-confidence interval windows, abnormal continuous segments obtained by clustering, and the detection result output table.

[0214] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for detecting deformation in continuous sections of shield tunnel segment joints, characterized in that, Includes the following steps: Extract the point cloud of the target joint and construct a local orthogonal reference coordinate system based on the main direction of the joint and the surface normal vectors of the segments on both sides of the joint; The smallest analysis unit is divided along the main direction of the seam. A point cloud quality field is constructed based on the point cloud density, local missing rate, balance of points on both sides, normal dispersion and fitting residual. The analysis window is expanded, merged or invalidated based on the point cloud quality field. Calculate the misalignment and opening within the effective window, and fuse the results of overlapping windows according to the window confidence level; Identify continuous sections of misalignment anomalies, opening anomalies, and compound anomalies based on the fusion deformation.

2. The method for detecting continuous deformation sections of shield tunnel segment joints according to claim 1, characterized in that, The extraction of the target joint point cloud and the construction of a local orthogonal reference coordinate system based on the main direction of the joint and the surface normal vectors of the segments on both sides of the joint include: S1: Use laser scanning equipment to acquire three-dimensional point cloud data of the joint area of ​​shield tunnel segments; S2: Perform outlier removal, voxel downsampling, normal vector estimation, and local smoothing on the three-dimensional point cloud data to obtain a standardized point cloud of the seam region; S3: Extract candidate seam point clouds based on the local geometric abrupt change features of the standardized point cloud of the seam region, and perform connectivity screening on the candidate seam point clouds to obtain the target seam point cloud. S4: Perform robust principal direction analysis on the target joint point cloud to obtain the joint principal direction axis; perform direction consistency correction and weighting on the average normal vectors of the segments on both sides of the joint, and orthogonalize the weighted direction vectors relative to the joint principal direction to obtain the misalignment direction axis; obtain the opening direction axis based on the cross product direction of the joint principal direction axis and the misalignment direction axis, thereby constructing a mutually orthogonal local reference coordinate system for the joint.

3. The method for detecting continuous deformation sections of shield tunnel segment joints according to claim 2, characterized in that, The process involves dividing the analysis window into minimum analysis units along the main direction of the seam, constructing a point cloud quality field based on point cloud density, local missing rate, balance of points on both sides, normal dispersion, and fitting residuals, and then expanding, merging, or invalidating the analysis window based on this point cloud quality field, including: S5: Divide the target seam point cloud into multiple seam minimum analysis units along the main direction of the seam, and calculate the point cloud density, local missing rate, balance of points on both sides of the seam, normal dispersion and local fitting residual of each seam minimum analysis unit. S6: Normalize and weight the point cloud density, local missing rate, point balance on both sides of the seam, normal dispersion, and local fitting residual to construct a point cloud quality field distributed along the main direction of the seam, and generate a quality-constrained seam analysis window based on the point cloud quality field; wherein, an initial window is generated with a preset initial window length; when the initial window does not meet the preset window quality conditions, the window is expanded along the main direction of the seam with the smallest analysis unit of the seam as the expansion step size; when the expanded window meets the preset window quality conditions, it is determined as a valid expanded window; when there are local low-quality units or local missing units between adjacent windows, and the merged window meets the preset window quality conditions, the adjacent windows are merged into a valid merged window; when the window reaches the preset maximum window length but still does not meet the preset window quality conditions, it is determined as an invalid window.

4. The method for detecting continuous deformation sections of shield tunnel segment joints according to claim 3, characterized in that, The calculation of misalignment and opening within the effective window, and the fusion of overlapping window results according to window confidence, includes: S7: Within each effective quality constraint joint analysis window, separate the point clouds of the pipe segments on both sides of the joint according to the joint local reference coordinate system, and establish local surface models on both sides of the joint respectively. S8: Under the local reference coordinate system of the joint, calculate the local misalignment based on the corresponding position difference of the local patch models on both sides of the joint on the misalignment direction axis, and calculate the local opening based on the projection distance of the feature points on both sides of the joint on the opening direction axis. S9: Calculate the window confidence based on the point cloud integrity, point balance on both sides of the joint, local patch fitting residual, normal consistency, and deformation continuity of adjacent effective windows for each effective mass constraint joint analysis window. S10: Remove windows with a confidence level lower than the preset confidence threshold, and perform confidence-weighted fusion of the local misalignment and local opening of multiple effective quality constraint joint analysis windows covering the same position in the main direction of the joint to obtain a fused misalignment sequence and a fused opening sequence distributed along the main direction of the joint.

5. The method for detecting continuous deformation sections of shield tunnel segment joints according to claim 4, characterized in that, The method of identifying continuous sections of misalignment anomalies, opening anomalies, and compound anomalies based on fusion deformation includes: S11: Based on the comparison results of the fused misalignment sequence and the fused opening sequence with the corresponding deformation threshold, the abnormal window is divided into misalignment abnormal window, opening abnormal window, or misalignment-opening composite abnormal window, and clustering is performed according to the spatial continuity condition and deformation continuity condition corresponding to different abnormal types; when there is a low confidence interval window between two abnormal candidate segments, and the length of the main direction of the joint corresponding to the low confidence interval window is less than the preset interval length threshold, and the abnormal candidate segments on both sides of the interval meet the corresponding deformation continuity condition, the two abnormal candidate segments are merged to obtain the joint deformation abnormal continuous segment; S12: Output the starting position, ending position, section length, maximum misalignment, average misalignment, maximum opening, average opening, section confidence level, and anomaly type of the continuous section with abnormal joint deformation, as the detection result of the continuous section with abnormal joint deformation of the shield tunnel segment.

6. The method for detecting continuous deformation sections of shield tunnel segment joints according to claim 2, characterized in that, In S4, the joint local reference coordinate system includes the joint principal direction axis, the misalignment direction axis, and the opening direction axis; The average normal vectors of the pipe segments on both sides of the joint are corrected for directional consistency so that they point to the same side. Based on the corrected normal vectors on both sides, the initial misalignment direction vector is calculated by weighting with the corresponding mass weights. Then, the initial misalignment direction vector is orthogonalized relative to the main direction axis of the joint to obtain the misalignment direction axis. The opening direction axis is determined by the cross product of the main direction axis of the joint and the misalignment direction axis; The coordinates of any point in the target seam point cloud under the local reference coordinate system of the seam are determined by the spatial coordinates of the point, the origin of the coordinate system, and the direction vectors of each coordinate axis.

7. The method for detecting continuous deformation sections of shield tunnel segment joints according to claim 3, characterized in that, In S5, the point cloud density is determined by the number of effective points within the smallest analysis unit of the seam, the unit length, and the bandwidth participating in the analysis on both sides of the seam. The local missing rate is determined by the ratio of the length of the seam direction covered by the effective point cloud within the smallest analysis unit of the seam to the length of the unit. The balance of the number of points on both sides of the joint is determined by the degree of difference in the number of effective points on the left and right sides of the joint within the unit; the smaller the difference, the higher the balance.

8. The method for detecting continuous deformation sections of shield tunnel segment joints according to claim 7, characterized in that, In S5, the normal dispersion is determined by the statistical value of the deviation between the local normal vector of each point in the smallest analysis unit of the joint and the average normal vector of the unit; The local fitting residual is determined by the statistical values ​​of the distances from each point within the smallest analysis unit of the seam to the local patch fitting model.

9. The method for detecting continuous deformation sections of shield tunnel segment joints according to claim 3, characterized in that, In S6, the point cloud quality value of each unit in the point cloud quality field is obtained by weighted summation of the normalized point cloud density, local missing rate, point balance, normal dispersion and local fitting residual. The normalization of each parameter adopts a threshold truncation method, and is normalized with preset reference values ​​respectively; When the number of valid points in a certain cell is lower than the preset minimum number of points, the point cloud quality value of that cell is set to zero and marked as a missing cell.

10. The method for detecting continuous deformation sections of shield tunnel segment joints according to claim 9, characterized in that, In S6, generating a mass constraint joint analysis window based on the point cloud mass field includes: Multiple initial windows are generated sequentially along the main direction of the seam with a preset initial window length and a sliding step size smaller than that length, with adjacent windows partially overlapping; The overall quality value of a window is determined by the ratio of the sum of the point cloud quality values ​​of the smallest analysis units of each seam contained therein to the number of units. An effective window must simultaneously meet the following requirements: the overall quality value is not lower than the preset quality threshold, the number of effective points is not lower than the preset minimum number of points, the local missing rate is not higher than the preset maximum allowable missing rate, and the balance of the number of points on both sides of the seam is not lower than the preset minimum allowable balance. If the initial window does not meet the conditions, it is gradually expanded with the smallest analysis unit of the seam as the step size until the conditions are met or the preset maximum window length is reached; the expanded window that meets the conditions is considered a valid window. When there are local low-quality or missing units between adjacent windows, but the merged window can meet the valid window conditions, the two are merged into one valid window; If the window reaches its maximum length but still does not meet the conditions, or if the missing rate within the window continues to exceed the limit, it is determined to be an invalid window and will not participate in subsequent deformation calculations.

Images