A data technology-based shield tunnel structure calculation system and calculation method

Abstract

The application discloses a kind of shield tunnel structure calculation system and calculation method based on data technology, specifically relates to engineering data processing field, for solving the problem that point cloud element and calculation grid lack consistency rule and lead to coordinate deviation, mapping drift and deformation interpretation inaccuracy, by establishing alignment dictionary under unified reference coordinate and time anchor point, distribution ring level identification primary key, record local origin and grid node correspondence;According to the semantic mapping of point cloud element by alignment dictionary, generate grid mapping table, mark ring segment and time anchor point;Tolerance verification and drift diagnosis are carried out on grid mapping table, calculate seam line fingerprint and closed communication, generate alignment decision coefficient and determine after comprehensive analysis;Translate the mapping table that passes into difference patch, only update change ring segment load and constraint;Read shape backcast quantity and perform consistency check, when not passed, reconstruct mapping table and patch according to conflict arc segment, when passed, solidify alignment dictionary and mapping table.

Description

Technical Field

[0001] This invention relates to the field of engineering data processing, and more specifically, to a shield tunnel structure calculation system and calculation method based on data technology. Background Technology

[0002] During the construction and operation of shield tunnels, the monitoring side acquires massive point clouds and morphological information through laser scanning and image acquisition. Through multi-location stitching and coordinate unification, geometric elements such as annular joints, axes, and component positions are segmented according to the tunnel segment boundaries. The structural calculation side establishes computational grids for annular segments and joints in a digital environment, updating loads and constraints based on operating conditions. The geometric and state quantities generated by monitoring must be accurately mapped to the corresponding annular segments and grid nodes, ensuring that on-site data can be stably transformed into executable computational inputs.

[0003] In the current process, there is a lack of machine-verifiable consistency rules between point cloud elements, ring segment identifiers and computational grids. The monitoring results have slight deviations from the computational model in terms of coordinate reference, time anchor point and ring segment positioning. When the acquisition time, equipment conditions or occlusion are inconsistent, the deviation is amplified, and the mapping relationship drifts or overlaps, causing observations such as deformation and leakage to fall in the wrong position, which in turn affects the interpretation of internal forces and deformation.

[0004] To address the aforementioned problems, a technical solution is provided. Summary of the Invention

[0005] To overcome the shortcomings of the prior art, embodiments of the present invention provide a shield tunnel structure calculation system and method based on data technology. By establishing the correspondence between ring segment identifiers and grids under unified coordinates and time, performing semantic mapping of point cloud elements, verifying drift and diagnosing deviations, generating targeted update patches, checking the consistency of solution results and solidifying data, the system solves the problems of point cloud and grid mapping drift, coordinate deviation and inaccurate deformation interpretation in the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A method for calculating the structure of a shield tunnel based on data technology includes the following steps:

[0008] S1: Establish an alignment dictionary under a unified reference coordinate and a unified time anchor point, assign a ring-level identifier primary key to each ring segment, and record the correspondence between the local origin and the grid nodes;

[0009] S2: Perform semantic mapping on point cloud features based on the alignment dictionary, generate a mesh mapping table, and label the ring segments and time anchor points;

[0010] S3: Perform tolerance verification and drift diagnosis on the mesh mapping table, calculate the suture fingerprint matching degree and circumferential closure connectivity, generate alignment adjudication coefficients through comprehensive analysis of segment cross evidence weaving, and determine pass, correction or rollback according to the alignment adjudication coefficients;

[0011] S4: Translate the mesh mapping table that has passed the judgment into the solution input differential patch, only update the load and constraints of the changed loop, and keep the unchanged loop unchanged;

[0012] S5: Read the shape feedback amount of the solution result and perform consistency check within the unified reference coordinate. If it fails, locate the conflict arc segment according to the alignment adjudication coefficient and rebuild the mesh mapping table and differential patch. If it passes, solidify the alignment dictionary and mesh mapping table.

[0013] In a preferred embodiment, in step S1, a non-repeatable ring-level identifier primary key is generated for each ring segment. The center of the cylindrical surface fitted by the ring seam point set is used as the local origin. The local three directions are determined by the ring surface normal and the ring tangent. Rigid body transformation is used to project the mesh nodes onto the local three directions to obtain the local coordinates of the nodes. The one-to-one correspondence between the ring-level identifier primary key and the local origin to the mesh nodes is recorded to form a searchable key value index and coordinate transformation relationship.

[0014] In a preferred embodiment, step S2 reads the alignment dictionary, extracts point cloud features within a unified coordinate system, generates line elements and surface elements according to the direction of the circumferential seam and the component boundary, performs a restricted search only along the circumferential and normal directions, positions the line elements and surface elements within the ring-level identifier primary key and node local coordinate domain, generates a mesh mapping table, and writes four fields—ring-level identifier primary key, unified time anchor point, node local coordinates, and point cloud feature type—into the mapping record.

[0015] In a preferred embodiment, in step S3, the suture fingerprint matching calculation extracts the tortuous rhythm of the suture direction, the rhythm of the anchor hole appearance, and the rhythm of the boundary concavity and convexity from the point cloud, generates a fingerprint sequence with a fixed order, and standardizes the length and direction in the local three directions. The fingerprint sequence is then slid-matched along the circumferential direction at the grid boundary, allowing only circumferential translation and small normal alignment, and the coverage of continuous hit segments and interrupted segments is statistically analyzed.

[0016] In a preferred embodiment, in step S3, the circumferential closed connectivity calculation is based on the ring-level identifier primary key and the local coordinates of the nodes to construct a circumferential adjacency graph. Candidate closed loops are formed by the geometric adjacency of adjacent nodes, and backtracking and intersections are eliminated. High-density segments of the point cloud are selected as evidence pins to verify the candidate closed loops segment by segment, and the continuous coverage and empty segment ratio of the evidence pins in the circumferential direction are evaluated.

[0017] In a preferred embodiment, in step S3, when the alignment decision coefficient is higher than the pass threshold, it is marked as pass; when it is in the middle range, a correction signal is output; and when it is lower than the backoff threshold, a backoff signal is output.

[0018] In a preferred embodiment, in step S4, the mapping marked as passed is translated to generate a solution input difference patch. The patch is described by ring-level identifier primary key and node local coordinates and point cloud feature type to indicate the trigger range. Load update and constraint update instructions are written only for the changed ring segment. A source chain identifier and version identifier are written for each patch to form a traceable update sequence.

[0019] In a preferred embodiment, in step S5, the shape projection amount of the solution result is read, and the shape is projected back to the ring-level identifier primary key and node local coordinates in a unified reference coordinate system using coordinate transformation relationship. A comparison field is generated and a consistency check is performed. The comparison field and point cloud elements are checked for overlay of position and orientation in the same coordinate domain.

[0020] In a preferred embodiment, if the consistency check fails in step S5, the conflicting arc segment is located based on the alignment adjudication coefficient, the mesh mapping table and the solution input differential patch of the corresponding arc segment are reconstructed, and a recalculation is initiated. If the consistency check passes, a freeze mark and version identifier are written on the alignment dictionary and mesh mapping table, and the key-value index and adjudication record of the archive processing are archived.

[0021] A shield tunnel structure calculation system based on data technology, comprising:

[0022] Alignment dictionary construction unit: An alignment dictionary is established under a unified reference coordinate and a unified time anchor point. A ring-level identifier primary key is assigned to each ring segment, and the correspondence between the local origin and the grid node is recorded.

[0023] Semantic mapping generation unit: Performs semantic mapping on point cloud features based on the alignment dictionary, generates a mesh mapping table, and labels the ring segments and time anchor points;

[0024] Verification and diagnosis judgment unit: Performs tolerance verification and drift diagnosis on the grid mapping table, calculates the suture fingerprint matching degree and circumferential closure connectivity, generates alignment adjudication coefficients through comprehensive analysis of segment cross evidence weaving, and determines whether to pass, correct or revert according to the alignment adjudication coefficients;

[0025] Differential Patch Translation Unit: Translates the mesh mapping table that has passed the judgment into a solution input differential patch, only updating the loads and constraints of the changed loops, while keeping the unchanged loops unchanged;

[0026] Consistency check and solidification unit: Read the shape feedback amount of the solution result and perform consistency check within the unified reference coordinate. If it fails, locate the conflict arc segment according to the alignment adjudication coefficient and rebuild the mesh mapping table and differential patch. If it passes, solidify the alignment dictionary and mesh mapping table.

[0027] The technical effects and advantages of the shield tunnel structure calculation system and method based on data technology of this invention are as follows:

[0028] This invention establishes a closed-loop data processing and verification mechanism by creating an alignment dictionary under unified coordinates and time, performing semantic mapping of point cloud elements, verifying drift and diagnosing deviations, generating targeted update patches, and checking the consistency of solution results. This significantly improves the accuracy and stability of shield tunnel structure calculations. The precise correspondence between the ring-level identifier primary key and the grid nodes eliminates coordinate deviations; semantic mapping and verification diagnosis ensure the semantic consistency and topological integrity of the point cloud data and the calculation model; differential patching optimizes update efficiency; and the consistency verification and solidification mechanism further verifies and locks in reliable results. This avoids the cumulative effects of mapping drift and observation misalignment in traditional processes, achieves seamless conversion from monitoring data to calculation input, enhances the reliability of internal force and deformation interpretation, and provides efficient and stable technical support for tunnel construction and operation. Attached Figure Description

[0029] Figure 1 This is a flowchart illustrating a shield tunnel structure calculation method based on data technology according to the present invention.

[0030] Figure 2 This is a schematic diagram of the structure of a shield tunnel structure calculation system based on data technology according to the present invention. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] Example 1: Figure 1 This invention presents a method for calculating the structure of a shield tunnel based on data technology, comprising:

[0033] S1: Establish an alignment dictionary under a unified reference coordinate and a unified time anchor point, assign a ring-level identifier primary key to each ring segment, and record the correspondence between the local origin and the grid nodes.

[0034] S2: Perform semantic mapping on point cloud features based on the alignment dictionary, generate a mesh mapping table, and label the loop segments and time anchor points.

[0035] S3: Perform tolerance verification and drift diagnosis on the grid mapping table, calculate the suture fingerprint matching degree and circumferential closure connectivity, generate alignment adjudication coefficients through comprehensive analysis of segment cross evidence weaving, and determine whether to pass, correct or regress according to the alignment adjudication coefficients.

[0036] S4: Translate the mesh mapping table that has passed the judgment into a solution input differential patch, update only the loads and constraints of the changed loops, and keep the unchanged loops unchanged.

[0037] S5: Read the shape feedback amount of the solution result and perform consistency check within the unified reference coordinate. If it fails, locate the conflict arc segment according to the alignment adjudication coefficient and rebuild the mesh mapping table and differential patch. If it passes, solidify the alignment dictionary and mesh mapping table.

[0038] This invention provides a data-driven method for calculating the structure of a shield tunnel. Through a systematic data alignment and mapping mechanism, it achieves precise fusion of monitoring-side point cloud data and structural calculation-side mesh models. The method first constructs an alignment dictionary under unified reference coordinates and unified time anchor points, generating a unique ring-level identifier primary key for each ring segment and recording the correspondence between the local origin and mesh nodes to establish a reliable coordinate transformation basis. Subsequently, semantic mapping is performed on the point cloud elements based on this alignment dictionary, generating a mesh mapping table annotating ring segments and time anchor points, ensuring the uniqueness and traceability of mapping entries. Next, tolerance verification and drift diagnosis are performed on the mesh mapping table. By calculating the seam fingerprint matching degree and circumferential closure connectivity, and through machine learning comprehensive analysis, alignment adjudication coefficients are generated to determine the validity of the mapping and decide whether to pass, correct, or revert. For mappings that pass the judgment, they are translated into solution input differential patches, updating only the loads and constraints of the changed ring segments to minimize computational overhead. Finally, the morphological feedback quantity of the structural solution results is read, and a consistency check is performed within a unified reference coordinate system. If the check fails, the conflict is located and the relevant components are rebuilt based on the adjudication coefficient. If the check passes, the alignment dictionary and mesh mapping table are fixed, forming a complete process archive. This method, through a closed-loop logic of iterative verification and differential updates, solves the problems of coordinate deviation and mapping drift in existing processes, ensuring that monitoring data is stably converted into computational input, and improving the accuracy and efficiency of shield tunnel structural interpretation.

[0039] In shield tunnel structural calculations, the monitoring side acquires point cloud data through laser scanning and image acquisition, and extracts geometric elements such as annular joints and axes, while the structural calculation side constructs a segmental mesh model in a digital environment. The current process lacks consistency rules between point cloud elements and the computational mesh, leading to coordinate deviations and mapping drifts, which affect deformation interpretation.

[0040] The specific processing logic of step S1:

[0041] Step S1.1: Generate ring-level identifier primary key.

[0042] Based on the circumferential joint boundaries and component locations extracted from the monitoring side in the background technology, a unique identifier is first assigned to each ring segment. From the point cloud data under a unified reference coordinate system, a continuous set of circumferential joint points is selected as input, and a non-repeatable ring-level identifier primary key is generated through serialization encoding. Specifically, the starting point coordinates of the circumferential joint point set are combined with the ring segment number, and a hash function is used to calculate the primary key, ensuring that the identifier of each ring segment is unique and tamper-proof throughout the entire tunnel structure. This forms a unique identity marker for the ring segment, facilitating the tracking of the correspondence between monitoring data and the calculation model under a unified time anchor point, and avoiding ring segment positioning drift.

[0043] Step S1.2: Determine the local origin.

[0044] Using the ring-level identifier primary key associated with the ring seam point set, a cylindrical surface is fitted to locate the local origin. Point cloud samples from the ring seam point set are selected, and the least squares method is used to fit the cylindrical surface model, where the axis of the cylindrical surface is determined by the average normal vector of the point cloud. The specific fitting process is as follows: Let the ring seam point set be a set... Each of them Let be a three-dimensional coordinate point; define the direction vector of the cylindrical surface axis as . The primary eigenvectors of the point set are obtained through principal component analysis and used as... This is done to ensure that the axis aligns with the direction of the circumferential seam; then, the center point of the cylindrical surface is calculated. As a local origin, its formula is: In the formula Refers to the coordinates of a single point in the circumferential joint point set. Refers to the vector of the axis direction. Refers to the center point to be optimized. The cylindrical radius parameter ensures that the fit focuses on the geometric center without introducing bias. Point cloud data is bound to the ring-level identifier primary key, and the local origin coordinates are output to define the local coordinate system. This establishes the geometric reference center for each ring segment, supporting subsequent coordinate transformations and providing stable anchor points for addressing deviation issues in tunnel ring segment deformation monitoring.

[0045] Step S1.3: Define the local three-dimensional plane.

[0046] Based on a defined local origin, the torus normal and tangential directions are calculated to construct a local three-dimensional coordinate system. First, the torus normal vector is calculated from the set of torus seam points. The average normal vector is obtained through local curvature analysis of the point set, ensuring... Perpendicular to the toroidal plane; secondly, select the circumferential tangential vector. It is derived from the positive cross product of the axis and normal of the cylindrical surface, that is... Finally, define the third axial vector. This completes the orthogonal coordinate system. (The formula contains...) Refers to the normal vector of the torus. The axial vectors are used to ensure that the three vectors are orthogonal to each other. Relying on the local origin and axis, local three-dimensional basis vectors are output for mesh node projection, constructing a local coordinate framework specific to the loop segment, and laying the directional reference for the coordinate unification problem between the point cloud and the mesh.

[0047] Step S1.4: Perform grid node projection and record the corresponding relationships.

[0048] Rigid body transformation is used to project the computational mesh nodes onto the local three-dimensional plane, obtaining the local coordinates of the nodes. The specific transformation process is as follows: for each mesh node's global coordinates... By applying translation and rotation matrices, the coordinate system is transformed from a unified reference coordinate system to one with a local origin. A local coordinate system with the origin as its origin. ,in The rotation matrix is ​​formed by the local triaxial basis vectors. The orthogonal matrix formed ensures that the transformation involves only rigid body motion and does not introduce scaling. In this formula... Refers to the global coordinates of a grid node. This refers to the local coordinates of the projected nodes. Subsequently, a one-to-one correspondence is recorded between the ring-level identifier primary key and the local origin to each mesh node. The system associates data to form key-value indexes and coordinate transformation relationships. The generated searchable alignment dictionary provides accurate coordinate correspondences to address the amplified deviation between monitoring data and computational inputs, ensuring stable mapping relationships.

[0049] An alignment dictionary is established under a unified reference coordinate and a unified time anchor point. A unique ring-level identifier primary key is generated for each ring segment. The center of the cylindrical surface fitted by the ring seam point set is used as the local origin. The local three directions are determined by the ring surface normal and the ring tangent. Rigid body transformation is used to project the mesh nodes onto the local three directions to obtain the local coordinates of the nodes. The one-to-one correspondence between "ring-level identifier primary key, local origin to mesh node" is recorded to form a searchable key-value index and coordinate transformation relationship.

[0050] Step S1 successfully established a reliable correspondence between ring-level identifier primary keys and node local coordinates by building an alignment dictionary, providing a solid coordinate transformation foundation for the semantic mapping of point cloud elements. However, without further precise semantic mapping, the existing geometric element extraction cannot guarantee uniqueness and stability, and is prone to errors due to device occlusion or time differences, thereby weakening the reliability of the entire calculation process. Therefore, step S2 performs semantic mapping on point cloud elements based on the alignment dictionary, generating a grid mapping table of labeled ring segments and time anchor points. This not only ensures the unique source and destination of monitoring data, but also effectively suppresses the drift amplification effect through a restricted search mechanism, ultimately achieving a stable and efficient conversion of field data into computational input.

[0051] The specific processing logic of step S2:

[0052] Step S2.1: Read the alignment dictionary.

[0053] Based on the alignment dictionary formed in step S1, its key-value index and coordinate transformation relationship are first loaded, including the ring-level identifier primary key, local origin, and node local coordinates. Specific operations involve parsing dictionary records to ensure data availability under a unified reference coordinate and a unified time anchor point, and retrieving the corresponding entry for each ring segment through the index.

[0054] Step S2.2: Extract point cloud features and generate line elements and surface elements.

[0055] Using an alignment dictionary, the orientation of the circumferential seam and the boundary features of the component are extracted from point cloud data within a unified reference coordinate system. Specifically, an edge detection algorithm is applied to the point cloud point set to identify the circumferential seam curve, and then line elements (representing the edge of the circumferential seam) and surface elements (representing the surface of the segment) are generated according to the component boundary. Line element generation uses parametric curve fitting, such as Bézier curves, to approximate the circumferential seam point set; surface element generation uses triangulation to mesh the surface point set, ensuring that the features are consistent with the orientation of the circumferential seam. The extraction range is limited according to the alignment dictionary, and the line element and surface element datasets are output for constrained search. A geometric representation of the starting point cloud is established to provide a semantic basis for scenarios where feature deviations are magnified in the current process, avoiding mapping overlap.

[0056] Step S2.3: Perform a restricted search to locate line elements and surface elements.

[0057] Based on sub-line elements and polygon elements, a restricted search is performed only along the circumferential and normal directions to locate them within the "circumferential identifier primary key, node local coordinates" domain. For each line element or polygon element, the search space is restricted to the circumferential tangential and circumferential normal planes. The nearest neighbor matching method is used to find the corresponding point in the node local coordinates to achieve precise anchoring of features and provide a restricted constraint mechanism for drift caused by acquisition occlusion.

[0058] For example, the specific logic for finding the corresponding point in the local coordinates of a node using the nearest neighbor matching method can be as follows:

[0059] For each feature point on a line element or surface element, a spatial index (such as a KD-tree) of the grid points is first constructed within the local coordinate domain of the node to accelerate the query. Then, a nearest neighbor search is performed on the feature point coordinates, calculating the Euclidean distance between it and each candidate grid point, and adjusting it in conjunction with the cosine similarity of the normal vectors: the matching distance is the Euclidean norm of the feature point coordinates and the grid point coordinates multiplied by (1 minus the absolute dot product of the feature point normal vector and the grid point normal vector), where the Euclidean norm is obtained by subtracting the vector magnitude of the grid point coordinates from the feature point coordinates, and the absolute dot product is obtained by absolutizing the cosine value of the feature point normal vector and the grid point normal vector. The grid point with the smallest matching distance is selected as the corresponding point. If the distance exceeds a preset threshold, it is marked as no match to ensure that the anchoring considers the consistency of position and direction. This threshold is determined by analyzing the distance distribution of qualified samples and using kernel density estimation to determine the upper limit, such as taking the 95th percentile of the distribution to adapt to the variation of point cloud density.

[0060] Step S2.4: Generate a grid mapping table and write the fields.

[0061] Based on the positioning results, a grid mapping table is compiled and generated, with four fields written into each record: "ring-level identifier primary key, unified time anchor point, node local coordinates, and point cloud feature type". The compilation process involves verifying uniqueness: checking the exclusive correspondence between the source (point cloud feature) and destination (grid node) of each mapping entry; if a conflict is found, a backtracking search is performed. Fields are written to ensure the time anchor point is labeled with the acquisition time, and feature type classifications such as "ring seam element" or "tube segment element". Following the positioning output of step S2.3, a complete mapping table is generated for verification in step S3. After processing, an labeled mapping structure is formed, ensuring data uniqueness and traceability to address the issue of misplaced deformation observations.

[0062] Step S2, by reading the alignment dictionary, extracting and generating the geometric representation of point cloud features, performing constrained search localization, and compiling the mesh mapping table, achieves the semantic association between monitoring data and the computational grid, ensuring the uniqueness and stability of the mapping. It effectively bridges the correspondence between point cloud features and the ring-shaped mesh, providing a targeted solution to common problems in shield tunnels such as coordinate drift and time inconsistency. Finally, it outputs a fully annotated mesh mapping table, laying a reliable foundation for subsequent verification and computational input.

[0063] Steps S1 and S2, through the establishment of the alignment dictionary and the generation of the grid mapping table, have provided initial guarantees for data correspondence at the coordinate and semantic levels, improving the initial accuracy of the transformation. However, without in-depth verification of the mapping table, existing subtle drifts will still accumulate in the calculation, weakening the overall efficiency of the method and making it difficult to handle semantic biases caused by equipment conditions or occlusion. Therefore, step S3 introduces tolerance verification, discriminant parameter calculation, and machine learning-driven comprehensive analysis to generate alignment decision coefficients and make judgments. This not only accurately diagnoses drifts and ensures the semantic uniqueness and topological integrity of the mapping, but also achieves objective consistency without relying on subjective thresholds through an evidence weaving mechanism. Addressing the root cause of deviation amplification in the background technology, by establishing a machine-verifiable rule framework, the stability and interpretation accuracy of the computational input are significantly enhanced in tunnel dynamic monitoring, reducing potential fault risks and providing reliable diagnostic outputs for subsequent differential updates.

[0064] The specific processing logic of step S3:

[0065] Step S3.1: Perform tolerance verification and drift diagnosis item by item.

[0066] Based on the mesh mapping table generated in step S2, tolerance verification and drift diagnosis are performed on each record. Specific operations include traversing the records, quantifying whether the Euclidean position deviation between the local coordinates of the nodes and the point cloud feature type meets the preset tolerance, and assessing the circumferential offset trend by comparing the unified time anchor points of adjacent records. A diagnostic record set is established to provide a baseline filtering mechanism for mapping drift caused by variability in tunnel point cloud acquisition.

[0067] Step S3.2: Calculate the suture fingerprint matching degree.

[0068] Using a mesh mapping table, the zigzag rhythm of the suture orientation (quantified by curvature change sequence), the anchor hole occurrence rhythm (statistical hole position interval sequence), and the boundary concavity and convexity rhythm (recorded edge undulation sequence) are extracted from point cloud features. These are combined to generate a fingerprint sequence with a fixed order, and the length and direction are standardized by linear interpolation in the local three dimensions. Subsequently, the fingerprint sequence is slid-matched along the circumferential direction at the mesh boundary, allowing only circumferential translation and small normal alignment (limiting the alignment deviation to a preset percentage less than the mesh spacing). The coverage relationship between continuous hit segments (continuous matching parts of the sequence) and interrupted segments (mismatch gaps) is statistically analyzed. The specific calculation logic is as follows: the suture fingerprint matching degree is calculated as the matching coverage ratio of the total arc length of the fingerprint sequence multiplied by (1 minus the interruption ratio). The matching coverage ratio is obtained by integrating the hit indicator function along the sequence parameters and dividing by the total arc length, and the interruption ratio is obtained by integrating the interruption indicator function along the sequence parameters and dividing by the total arc length. This calculation principle quantifies the semantic strength of continuous matching through proportionalization, ensuring that the coverage relationship reflects geometric consistency. The suture fingerprint matching degree value is output for subsequent comprehensive analysis. After completion, the semantic consistency of the quantified annular gap geometry is verified to strengthen the validation of semantic issues that lead to amplified deviations in the current process.

[0069] For example, the length and direction can be standardized by linear interpolation in a local three-dimensional region as follows:

[0070] In a local three-dimensional coordinate system, the specific process of standardizing the length and direction of a fingerprint sequence through linear interpolation is as follows: First, determine the target standardized length. (For example, the average arc length of the ring segment is set to a fixed value, such as 100 units, to ensure the consistency of sequence comparisons); for the length of the original sequence (M-point set), calculate the interpolation step size Then, index each target point. ( arrive ), calculate the corresponding original position The standardized value is calculated using a linear interpolation formula. ,in for Return to original value, for The original values ​​are then used; simultaneously, for direction standardization, the same interpolation is applied to the normal and tangential components of the sequence to ensure that the sequence vector is normalized to a unit length in the local three directions (toroidal normal, toroidal tangential, and axial). The parameters of this process include the target length. Step length And the endpoint boundary processing of the original sequence (e.g., boundary extrapolation to the nearest value).

[0071] For example, the logic for obtaining a tortuous rhythm can be:

[0072] The line element point set of the annular seam direction is selected from the point cloud data. First, the local curvature of each point is calculated. The curvature value is obtained by fitting the quadratic curve of the adjacent point set to form a curvature sequence. Specifically, for uniformly sampled points on the line element, the rate of change of the tangent vector of each sampled point is calculated in sequence, that is, the angle between two adjacent tangent vectors divided by the arc length interval, to generate a zigzag rhythm sequence representing the bending mode, ensuring that the sequence captures the continuous turning features of the annular seam.

[0073] For example, the logic for obtaining the anchor hole rhythm can be:

[0074] Anchor hole locations are identified in point cloud features, and a set of hole coordinate points is extracted through density clustering or template matching. Subsequently, these holes are sorted along the direction of the circumferential seam, and the Euclidean distance or arc length interval between adjacent holes is calculated to form an interval sequence. This sequence quantifies the occurrence rhythm of anchor holes, reflecting the periodicity or non-uniformity of hole distribution, and is used for subsequent fingerprint generation.

[0075] For example, the logic for obtaining concave-convex rhythms can be:

[0076] Edge line elements are extracted from the point cloud of the circumferential seam boundary. The normal offset of each boundary point is calculated. The concavity and convexity properties are determined by comparing the distance from the point to the fitting plane (positive for convex and negative for concave). The sign sequence and amplitude sequence of these offsets are recorded in circumferential order to form the edge undulation rhythm and capture the wave-like change characteristics of the boundary geometry.

[0077] Step S3.3: Calculate the circumferential closed connectivity.

[0078] Based on the grid mapping table, a cyclic adjacency graph is constructed according to "ring-level identifier primary key, node local coordinates". Candidate closed loops are formed through geometric adjacency (the distance between adjacent nodes is less than the grid resolution), and backtracking (repeated node paths) and intersections (self-intersecting paths) are eliminated. Then, high-density segments of the point cloud (segments with density exceeding the average) are selected as evidence pin coverage segments. The candidate closed loops are verified segment by segment to evaluate the continuous coverage and empty segment ratio of the evidence pins in the cyclic direction. The specific calculation logic is as follows: the cyclic closure connectivity is calculated by multiplying the strength-weighted average ratio of the coverage segment by (1 minus the penalty-weighted average ratio of the empty segment). The strength-weighted average ratio is obtained by multiplying the verification strength of each segment within the number of coverage segments by the sum of the cyclic lengths and dividing by the sum of the cyclic lengths of each segment. The penalty-weighted average ratio is obtained by multiplying the empty penalty value of each segment within the number of empty segments (based on the proportion of empty segment length, ranging from 0 to 1) by the sum of the cyclic lengths and dividing by the sum of the cyclic lengths of each segment. This operation evaluates the topological support integrity through weighted ratios to ensure the balanced quantification of coverage and emptiness. Outputs cyclic closure connectivity values ​​for use as machine learning input. Provides observation-supported evaluation and enhances integrity checks for topological problems with overlapping maps.

[0079] Step S3.4: Perform machine learning synthesis analysis of segment cross-evidence weaving.

[0080] The normalized suture fingerprint matching degree and circumferential closure connectivity in steps S3.2 and S3.3 are analyzed as follows: The ring segment is uniformly divided into several arc segments (e.g., each segment covers 1 / 8 of the ring circumference). Matching evidence (local matching degree sub-value) and connectivity evidence (local connectivity degree sub-value) are read on each arc segment. The coverage pattern (continuous high value sequence) and conflict pattern (alternating low value sequence) of the archived qualified samples are learned through a sequence model (such as a long short-term memory network). The order relationship between arc segment coverage continuity (rank order of pattern matching score sequence) and conflict density (deviation peak density) is calculated. The specific calculation logic is as follows: the alignment adjudication coefficient is calculated as the geometric mean of (coverage rank order multiplied by 1 minus conflict density) of all arc segments. The geometric mean is obtained by multiplying the product of the product terms of each arc segment within the number of arc segments and taking the root of the number of arc segments. The principle is to balance the order relationship through product aggregation to ensure the comprehensive objectivity of evidence weaving. Finally, the alignment adjudication coefficient is generated for judgment.

[0081] Regarding the specific processing steps and parameter settings for using machine learning, the following methods can be used:

[0082] First, feature engineering is performed: the ring segment is divided into 8 equally divided arc segments (each arc segment covers 1 / 8 of the ring circumference), and concordance evidence (local suture fingerprint concordance sub-value sequence) and connectivity evidence (local circumferential closure connectivity sub-value sequence) are extracted from each arc segment. These sub-value sequences are used as input feature vectors with a dimension of 16 (one pair of evidence values ​​for each arc segment); the feature vectors are further normalized to ensure that the value range is between [0,1].

[0083] The training data comes from archived qualified samples, with a total of 1,000 historical tunnel segment samples collected, of which qualified samples account for 70%. Each sample contains arc segment evidence sequence, coverage pattern label (continuous high value sequence features) and conflict pattern label (alternating low value sequence features). Features are generated manually or through simulation to ensure that the coverage pattern reflects the distribution of highly consistent arc segments and the conflict pattern captures areas of dense deviation.

[0084] The Long Short-Term Memory (LSTM) network model is adopted, with an input layer (16 dimensions), two hidden layers (64 units each, using the ReLU activation function to capture non-linear sequence relationships), and an output layer (2 dimensions, used to predict coverage rank and conflict density). The model has approximately 10,000 parameters, uses the Adam optimizer (learning rate 0.001), the loss function is mean squared error (MSE), the training strategy is batch gradient descent (batch size 32), the training epochs are 100, and an early stopping mechanism (patience value 10) is adopted to prevent overfitting.

[0085] The order relationship between arc coverage continuity and conflict density is calculated as follows: the coverage rank is obtained by sorting the average value of arc evidence (high value first, normalized to [0,1]), and the conflict density is normalized by counting the peak value of evidence deviation within the arc (deviation is defined as the number of times the evidence value is lower than the sample mean). The alignment adjudication coefficient is calculated by geometric mean, which is the root of the number of arcs opened by the product of (coverage rank multiplied by 1 minus conflict density) of all arcs. This mathematical logic balances the contribution of each arc, ensures the global consistency quantification of the order relationship, and avoids the dominance of a single arc.

[0086] Step S3.5: Make a judgment based on the alignment decision coefficient.

[0087] Based on the alignment decision coefficient, classification is performed: if the alignment decision coefficient is higher than the pass threshold, it is marked as pass; if it is in the middle range, a correction signal is output, requiring the output of conflict arc segments (a list of high-density arc segments) and the initiation of a single-item correction (local remapping) signal; if it is lower than the backoff threshold, a backoff signal is output, determining a backoff to the loop repositioning signal. The specific pass threshold is determined by statistically analyzing the distribution of the alignment decision coefficients of archived samples, for example, using the quantile method, taking the upper 75th quantile as the lower limit of the pass threshold, the 50th to 75th quantile as the middle range, and the lower 50th quantile as the upper limit of the backoff threshold, to adapt to tunnel data variations. A decision label is generated for translation in step S4. After processing, mapping classification is implemented, providing an executable path for drift diagnosis.

[0088] Step S3 performs a comprehensive diagnosis of the mesh mapping table through tolerance verification, discriminant parameter calculation, machine learning analysis, and coefficient determination, ensuring machine-verifiable consistency between semantics and topology. It accurately addresses coordinate drift and deviation amplification issues, provides an objective judgment framework for inconsistent data acquisition in shield tunnels, and ultimately outputs labeled results and conflict details, providing an analytical basis for optimizing computational inputs.

[0089] Step S3, through dictionary construction, semantic mapping generation, and verification diagnosis, provides a foundation for coordinate unification, unique correspondence, and machine judgment of monitoring data, ensuring the initial reliability of the input. However, if the mapping is not efficiently translated into computational input, existing updates of changed segments may still be inefficient or require global recalculation, wasting resources and delaying interpretation. Therefore, step S4 translates the mesh mapping table that has passed the judgment, generating differential patches for the solution input, updating loads and constraints only for changed segments, and adding tracking markers. This not only optimizes computational efficiency and ensures the stability of unchanged segments but also achieves the traceability of the update sequence. The implementation of this step solves the root cause of observational errors, significantly improves the accuracy and response speed of structural calculations under dynamic tunnel conditions, reduces the risk of full model overload, and provides efficient input for final consistency verification, supporting the reliability of engineering decisions.

[0090] The specific processing logic of step S4:

[0091] Step S4.1: Select the mapping record marked.

[0092] From the mesh mapping table marked as passed in step S3, all records that meet the alignment decision coefficient are filtered out, including "ring-level identifier primary key, unified time anchor point, node local coordinates, and point cloud feature type". The specific filtering process involves traversing the table entries and retaining only those that have passed the decision to ensure the reliability of the translation basis, form a refined subset of mappings, and avoid invalid update propagation.

[0093] Step S4.2: Determine the scope of the changed ring segment.

[0094] Using selected mapping records, the point cloud feature types under the current unified time anchor point are compared with previous archived versions to identify segments where geometric or state changes have occurred. Specifically, node-by-node comparison is used: if the feature types associated with the local coordinates of a node show deformation or leakage differences, the corresponding ring-level identifier primary key is marked as a changed segment. The output list of changed segments is used to limit the patch scope, precisely define the update trigger area, and optimize the calculation of deviation impact judgment issues in the current process.

[0095] Step S4.3: Generate the solution input difference patch and write the update command.

[0096] Based on the defined changed segments, the patch trigger range is described by "segment-level identifier primary key, node local coordinates, and point cloud feature type". Load update (adjusting force and torque based on deformation) and constraint update (modifying boundary conditions based on leakage features) instructions are compiled only for these segments; unchanged segments retain their original computational mesh. The specific compilation process involves generating instruction pairs for each changed node, ensuring that the patch only incrementally modifies the data. A differential patch file is output for structural solution input, implementing a differential update mechanism and providing efficient load constraint adjustment for drift caused by inconsistent data acquisition.

[0097] Step S4.4: Write the source chain identifier and version identifier.

[0098] For each differential patch generated from the solution input, a source chain identifier (linked to the mesh mapping record in step S2) and a version identifier (corresponding to the alignment dictionary version in step S1) are appended to the patch record, forming a chained update sequence. Key-value pairs are used to embed the identifiers, ensuring traceability throughout the entire chain from monitoring to computation. A final patch with the identifier is generated for traceable execution, establishing an update audit path and enhancing data stability and version control.

[0099] Step S4 transforms the verification results into efficient calculation input by selecting the mapping, determining the scope of change, generating differential patches, and adding identifiers, ensuring targeted updates only for changed segments. This effectively solves the inefficiency of full-domain recalculation and provides a traceable incremental mechanism for scenarios of mapping drift in shield tunnel operation. The final output is a versioned patch, providing an optimized foundation for subsequent form re-implementation verification.

[0100] Step S4, through alignment of dictionaries, mapping tables, diagnostic judgments, and differential patches, has achieved precise transformation and incremental updates of data from monitoring to solution, ensuring initial optimization of computational efficiency. However, without final consistency verification, the existing solution results may not match the on-site point cloud, weakening the closed-loop reliability of the overall method and failing to correct residual deviations in a timely manner. Therefore, step S5 reads the morphological re-projection of the solution results, performs coordinate re-projection, comparison generation, and verification, and reconstructs or solidifies the data based on the results. This not only verifies the geometric consistency of the computational output but also achieves iterative optimization and archiving traceability through conflict localization. A closed-loop feedback mechanism is established to address the consequences of incorrect observations, significantly improving interpretation accuracy and data stability in the dynamic environment of the tunnel, preventing potential misjudgment risks, and providing auditable processing records for long-term operation.

[0101] The specific processing logic of step S5:

[0102] Step S5.1: Read the shape return amount of the solution result.

[0103] Based on the structural calculation output after executing the differential patch of the solver input generated in step S4, the shape projection data, including deformation displacement and stress distribution data, is loaded first. The loading process involves parsing the output file of the computational mesh to ensure data integrity under a unified reference coordinate system. This forms an operable dataset of the shape projection data.

[0104] Step S5.2: Project the shape back to the local coordinates of the node.

[0105] Using the read morphological projection data, and applying the coordinate transformation relationship established in step S1, the projection data is projected from the unified reference coordinates to the "ring-level identifier primary key, node local coordinates" domain. The projection employs an inverse transformation operation: a deformation vector is applied to each grid node, and the data is mapped to the local three-dimensional coordinates through rotation and translation inverse matrices, ensuring alignment between the projection data and the origin cloud domain. Based on the projection data and the transformation relationship in step S1, the projection data in the local coordinates is output for comparison field construction. Upon completion, coordinate unification of the solution results is achieved, providing projection anchoring to address the deviation amplification issue in the current process.

[0106] Step S5.3: Generate a contrast field.

[0107] Based on the morphological back-projection data after sub-projection, point cloud elements are superimposed within the same coordinate domain to generate a contrast field. Specifically, this generation involves gridded interpolation: mapping the back-projection amount to the point cloud element type (e.g., circumferential seam location) on the local coordinates of the nodes to form a deviation field map, ensuring visual quantification of position and orientation. Relying on the local back-projection output, a contrast field dataset is generated for input verification, establishing a deviation visualization structure, and enhancing inspection preparation for cases of misaligned leakage observations.

[0108] Step S5.4: Perform a consistency check.

[0109] Based on the generated comparison field, position overlay (coordinate deviation calculation) and orientation overlay (cosine similarity of normal vectors) checks are performed between the comparison field and the point cloud features. The specific calculation logic is as follows: the consistency error is calculated as the maximum value of the position norm deviation among all corresponding points plus (1 minus the dot product of the orientation vectors). The position norm deviation is obtained by subtracting the Euclidean norm of the corresponding point cloud feature's local coordinate position from the local coordinate position of the re-projection point. The orientation vector dot product is obtained by the cosine similarity between the orientation direction vector of the re-projection point and the orientation direction vector of the point cloud feature. This calculation principle quantifies overlay inconsistencies through maximum deviation, ensuring that the error reflects the overall geometric differences. The judgment criterion is: if the consistency error is less than the verification threshold, the verification passes; otherwise, the verification fails. The verification threshold is determined using a quantile method by simulating the error distribution of qualified tunnel samples, for example, taking the upper 95th quantile as the upper limit of the threshold to accommodate the variability of different ring segments. The verification result (pass or fail) is output for subsequent decision-making.

[0110] Step S5.5: Reconstruct or solidify based on the verification results.

[0111] If the consistency check fails, based on the conflicting arcs output in step S3, the mesh mapping table is rebuilt for the corresponding arc (step S2 is re-executed for localization) and the input difference patch is solved (step S4 is recompiled for updating), and a structural recalculation is initiated. If the check succeeds, a freeze marker (unmodifiable label) and version identifier are written to the alignment dictionary and mesh mapping table, and the key-value index and adjudication record are archived. Specific reconstruction is limited to conflicting arcs to avoid global overload; solidification is accomplished through embedded metadata. This step relies on the check output of step S5.4 and the conflict information of step S3 to generate the final processing result for method closure. After processing, iterative correction or data solidification is implemented to ensure the stability of the computational closure loop in the case of drift caused by acquisition occlusion.

[0112] Step S5, through reading the backfeed amount, coordinate projection, comparison generation, verification execution, and result processing, achieves the final verification and optimization of the solution output and monitoring data. It provides a deviation correction and archiving mechanism for inconsistencies in mapping within the shield tunnel, ultimately outputting solidified data or recalculation instructions, providing comprehensive assurance for the accuracy of structural interpretation.

[0113] Example 2: Figure 2 This invention provides a data technology-based shield tunnel structure calculation system, comprising:

[0114] Aligned dictionary building blocks:

[0115] An alignment dictionary is established under a unified reference coordinate and a unified time anchor point. Each ring segment is assigned a ring-level identifier primary key, and the correspondence between the local origin and the grid nodes is recorded.

[0116] Semantic mapping generation unit:

[0117] Semantic mapping is performed on point cloud features based on the alignment dictionary to generate a mesh mapping table, and ring segments and time anchors are labeled.

[0118] Verification and diagnostic judgment unit:

[0119] Tolerance verification and drift diagnosis are performed on the grid mapping table, the suture fingerprint matching degree and circumferential closure connectivity are calculated, and the alignment adjudication coefficient is generated through comprehensive analysis of the segment cross evidence weaving. The alignment adjudication coefficient is used to determine whether to pass, correct or revert.

[0120] Differential patch translation unit:

[0121] The mesh mapping table that passes the judgment is translated into a solution input differential patch, and only the loads and constraints of the changed loops are updated, while the unchanged loops remain unchanged.

[0122] Consistency verification and solidification unit:

[0123] The shape feedback quantity of the solution result is read and a consistency check is performed within a unified reference coordinate. If it fails, the conflicting arc segment is located based on the alignment adjudication coefficient and the mesh mapping table and differential patch are rebuilt. If it passes, the alignment dictionary and mesh mapping table are fixed.

[0124] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0125] It should be noted that the system of the present invention can be deployed on the device itself to realize embedded applications, or it can run on a PC or other terminal with a user interface, thereby meeting a variety of hardware environments and usage requirements.

[0126] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0127] It should be noted that, in this document, the use of relational terms such as "first" and "second" is merely to distinguish one entity or operation from another, and does not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0128] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for calculating the structure of a shield tunnel based on data technology, characterized in that, Including the following steps: S1: Establish an alignment dictionary under a unified reference coordinate and a unified time anchor point, generate a non-repeating ring-level identifier primary key for each ring segment, take the center of the cylindrical surface fitted by the ring seam point set as the local origin, determine the local three directions with the ring surface normal and the ring tangent, use rigid body transformation to project the mesh nodes onto the local three directions to obtain the local coordinates of the nodes, record the one-to-one correspondence between the ring-level identifier primary key and the local origin to the mesh node, and form a searchable key value index and coordinate transformation relationship; S2: Perform semantic mapping on point cloud features based on the alignment dictionary, generate a mesh mapping table, and label the ring segments and time anchor points; S3: Perform tolerance verification and drift diagnosis on the mesh mapping table, calculate suture fingerprint matching degree and circumferential closure connectivity, generate alignment decision coefficients through comprehensive analysis of segment cross evidence weaving, and determine pass, correction or backtracking according to the alignment decision coefficients; Among them, the suture fingerprint matching degree calculation extracts the tortuous rhythm of the suture direction, the rhythm of anchor hole appearance and the rhythm of boundary concavity and convexity from the point cloud, generates a fingerprint sequence with a fixed order, and standardizes the length and direction in the local three directions. The fingerprint sequence is slid-matched along the circumferential direction at the mesh boundary, allowing only circumferential translation and slight normal contact, and statistically analyzes the coverage relationship between continuous hit segments and interrupted segments; The circumferential closure connectivity calculation constructs a circumferential adjacency graph based on the ring-level identifier primary key and the local coordinates of the nodes, forms candidate closed loops by the geometric adjacency of adjacent nodes and eliminates backtracking and crosses, selects high-density segments of the point cloud as evidence pins to verify the candidate closed loops segment by segment, and evaluates the continuous coverage and empty segment ratio of the evidence pins in the circumferential direction; S4: Translate the mesh mapping table that has passed the judgment into the solution input differential patch, only update the load and constraints of the changed loop, and keep the unchanged loop unchanged; S5: Read the shape feedback amount of the solution result and perform consistency check within the unified reference coordinate. If it fails, locate the conflict arc segment according to the alignment adjudication coefficient and rebuild the mesh mapping table and differential patch. If it passes, solidify the alignment dictionary and mesh mapping table.

2. The method for calculating the structure of a shield tunnel based on data technology according to claim 1, characterized in that: In step S2, the alignment dictionary is read, point cloud features are extracted within the unified coordinate system, line elements and surface elements are generated according to the direction of the circumferential seam and the boundary of the component, a restricted search is performed only along the circumferential and normal directions, the line elements and surface elements are located within the ring-level identifier primary key and node local coordinate domain, a mesh mapping table is generated, and four fields are written into the mapping record: ring-level identifier primary key, unified time anchor point, node local coordinates, and point cloud feature type.

3. The method for calculating the structure of a shield tunnel based on data technology according to claim 1, characterized in that: In step S3, if the alignment decision coefficient is higher than the pass threshold, it is marked as pass; if it is in the middle range, a correction signal is output; if it is lower than the backoff threshold, a backoff signal is output.

4. The method for calculating the structure of a shield tunnel based on data technology according to claim 3, characterized in that: In step S4, the mapping marked as passed is translated to generate the solution input difference patch. The patch is described by the ring-level identifier primary key, node local coordinates and point cloud feature type to describe the trigger range. Load update and constraint update instructions are written only for the changed ring segment. The source chain identifier and version identifier are written for each patch to form a traceable update sequence.

5. The method for calculating the structure of a shield tunnel based on data technology according to claim 4, characterized in that: In step S5, the shape projection amount of the solution result is read, and the shape is projected back to the ring-level identifier primary key and node local coordinates in the unified reference coordinates using coordinate transformation relationship. A comparison field is generated and a consistency check is performed. The comparison field and point cloud elements are checked for overlay of position and orientation in the same coordinate domain.

6. The method for calculating the structure of a shield tunnel based on data technology according to claim 5, characterized in that: If the consistency check fails in step S5, the conflicting arc segment is located based on the alignment adjudication coefficient, the mesh mapping table and the solution input difference patch of the corresponding arc segment are reconstructed, and a recalculation is initiated. If the consistency check passes, a freeze mark and version identifier are written on the alignment dictionary and mesh mapping table, and the key-value index and adjudication record of the archive processing are archived.

7. A shield tunnel structure calculation system based on data technology, used to implement the shield tunnel structure calculation method based on data technology as described in any one of claims 1-6, characterized in that, include: Alignment dictionary construction unit: An alignment dictionary is established under a unified reference coordinate and a unified time anchor point. A ring-level identifier primary key is assigned to each ring segment, and the correspondence between the local origin and the grid node is recorded. Semantic mapping generation unit: Performs semantic mapping on point cloud features based on the alignment dictionary, generates a mesh mapping table, and labels the ring segments and time anchor points; Verification and diagnosis judgment unit: Performs tolerance verification and drift diagnosis on the grid mapping table, calculates the suture fingerprint matching degree and circumferential closure connectivity, generates alignment adjudication coefficients through comprehensive analysis of segment cross evidence weaving, and determines whether to pass, correct or revert according to the alignment adjudication coefficients; Differential Patch Translation Unit: Translates the mesh mapping table that has passed the judgment into a solution input differential patch, only updating the loads and constraints of the changed loops, while keeping the unchanged loops unchanged; Consistency check and solidification unit: Read the shape feedback amount of the solution result and perform consistency check within the unified reference coordinate. If it fails, locate the conflict arc segment according to the alignment adjudication coefficient and rebuild the mesh mapping table and differential patch. If it passes, solidify the alignment dictionary and mesh mapping table.

Images