A power transmission and transformation engineering construction simulation method based on digital twinning

Abstract

This invention discloses a simulation method for power transmission and transformation engineering construction based on digital twins, specifically relating to the field of power transmission and transformation construction. It addresses the problem of unstable simulation and release decisions caused by uncertain boundaries of underground objects due to data discrepancies during power transmission and transformation construction. By constructing an evidence layer according to its source and establishing a unified identifier and spatial benchmark for underground objects, a credible corridor range with source labels is generated, and construction procedures are constructed as action envelopes. Combined with test pit excavation verification events to drive local alignment updates and conflict decomposition of corridors, a constraint trigger chain is deduced under the multi-corridor assumption, forming a minimum set of exploration closed-loop points. During the release phase, freeze and release conditions are set for actions reaching uncertain corridors, and the corridor version is re-deduced and written back after taking effect. Finally, the digital twin version chain is archived to support reproduction.

Description

Technical Field

[0001] This invention relates to the field of power transmission and transformation construction, and more specifically, to a simulation method for power transmission and transformation engineering construction based on digital twins. Background Technology

[0002] In the substation foundation pit excavation, cable trench and grounding grid construction phases of power transmission and transformation projects, existing underground pipelines and historical structures are often located nearby. Constrained by construction schedules and power outage / restoration arrangements, it is necessary to verify the location and impact range of underground objects before construction and organize machinery movement, excavation progress, and support layout accordingly. Current practices typically integrate design drawings, as-built data, geophysical exploration interpretation results, and limited test pit exposure information, and conduct construction simulations and risk checks in BIM / GIS or 3D scenes. However, due to inconsistencies in data accuracy, coordinate benchmarks, and interpretation standards from different sources, and the adjustments to temporary facilities and excavation boundaries during the process, underground objects are often represented in a single version or static line position, making it difficult to cover uncertain areas and changing links.

[0003] In the above scenario, inconsistencies in multi-source evidence regarding underground objects can lead to reversals in the feasibility conclusions of the same action sequence under different assumptions, resulting in unstable inferences. The mechanism is that underground location deviations and uncertain intervals are not uniformly and traceably expressed and evolve continuously with verification events. Furthermore, the selection of test pits or exploratory excavation points often relies on experience, making it difficult to quantify their marginal contribution to the convergence of uncertain intervals. When there is a lack of unified citation identifiers and version traceability, the source of conflict and the scope of impact are difficult to clarify. Excavation and support decisions may be based on outdated or contradictory underground facts, thereby causing risks of mis-excavation, work stoppages, rework and reinforcement, and damage to surrounding pipelines.

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

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a construction simulation method for power transmission and transformation engineering based on digital twins. This method constructs an evidence layer according to the source and establishes a unified identifier and spatial benchmark for underground objects. It generates a credible corridor range with source labels and constructs construction procedures as action envelopes. Combined with test pit excavation verification events to drive local alignment updates and conflict decomposition of corridors, it deduces constraint trigger chains under the multi-corridor assumption and forms a minimum set of exploration closed-loop points. During the release phase, it sets freeze and release conditions for actions that reach uncertain corridors and re-deduces and writes back after the corridor version takes effect. Finally, it archives the twin version chain to support reproduction, thereby solving the problems mentioned in the background art.

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

[0007] S1: Construct evidence layers according to source, establish unified identifiers for underground objects and bind spatial reference benchmarks, and generate a credible corridor range volume containing source labels; construct the construction action sequence into an action envelope and establish a reference mapping;

[0008] S2: Map the exposure results of the verification points to a unified identifier, constrain the trusted corridor range to perform local alignment updates to form a new version of the corridor; merge the inconsistencies in the evidence layer into four types of conflicts and label the influence domain;

[0009] S3: Derive the constraint triggering chain on the set of credible corridor assumptions and output the feasibility label, construct the stability index and the verification value index and input them into the verification priority coefficient generator, output the minimum set of exploration closed loop points and complete the action set splitting;

[0010] S4: Read the gating result and verification priority coefficient, freeze the set of uncertain corridor actions and use verification completion and new corridor version effectiveness as release conditions, allow access without touching the alternative action set and archive the twin version chain.

[0011] Furthermore, the evidence layers are composed of the design drawing evidence layer, the completion and operation and maintenance evidence layer, the exploration and interpretation evidence layer, and the historical exposure evidence layer. The construction survey control network is selected as the spatial reference benchmark. The rigid body is aligned with the corresponding control points, and outliers are eliminated to complete the geometric normalization of the evidence layers. A mapping table between the unified identifier of underground objects and the evidence layers is established.

[0012] Furthermore, based on the unified identifier, the center line is extracted from each evidence layer, a tubular credible corridor range volume is constructed, and the source label is registered. Based on the construction action sequence, action labels and action envelopes are generated. The reach indication and separation distance of the action envelope and the credible corridor range volume are calculated to form a reference mapping table, and the credible corridor range volume library and action envelope library are output.

[0013] Furthermore, the system receives verification events and registers verification points, exposure point sets, measurement method identifiers, and measurement resolution scales. Verification points are expressed using spatial reference benchmarks. Based on the trusted corridor range library, point inclusion retrieval and nearest distance retrieval are performed. Combined with the mapping table from the evidence layer to the unified identifier, the unified identifier is locked.

[0014] Furthermore, based on the unified identifier and exposure point set, the local alignment update rules are executed to update the layered centerline and radius function, and a new version of the corridor is generated. Based on the consistency of residual direction and the difference in orientation topology, conflict classification is completed, and the affected spatial domain is generated. The affected action reference domain is filtered according to the reference mapping table and written into the version chain.

[0015] Furthermore, based on the corridor version chain, the hierarchical corridor range set is read, and a set of credible corridor hypotheses is generated according to the source label. For the action sequence, the action envelope library is read, and the intersection and nearest distance determination of the action envelope with each hypothetical corridor are performed to form a constraint trigger chain. The feasibility label and first trigger record under each hypothesis are output.

[0016] Furthermore, the feasibility flip ratio is obtained by comparing the percentage difference between the statistical baseline hypothesis and the other hypotheses. A set of sensitive sections is generated on the construction layout line based on the first trigger record. A set of candidate verification points is generated around the sensitive sections by anchoring at the minimum net distance. The expected range reduction is estimated by calling the local alignment update rule. The reduction density of the corridor is obtained by combining the verification workload to characterize the volume.

[0017] Furthermore, based on the stability gating threshold, the feasible flip ratio is gating and screening is performed. A verification priority coefficient generator with monotonic constraints is used to map the feasible flip ratio and the corridor benefit density to verification priority coefficients and output the sorting. The minimum exploration closed-loop point set is formed according to the coverage rule. The set of actions that reach uncertain corridors and the set of actions that do not reach alternative corridors are split according to the reference mapping table.

[0018] Furthermore, the stability gating results and the minimum exploration closed-loop point set are read to generate a release unit record; the release unit record is bound to the action identifier, the unified identifier set of the reach source, and associated with the effective version index of the corridor version chain; for the reach of uncertain corridor action set, freeze judgment and release judgment are registered; the release judgment corresponds to the verification event completion status and the new version of the corridor effective status.

[0019] Furthermore, for actions that do not trigger alternative action sets, release is performed, and the execution window and execution trajectory are archived; for actions that trigger uncertain corridor action sets, release is performed and the execution window and execution trajectory are archived when the release condition is met; after the verification event is completed, the new version of the corridor is triggered to take effect, the affected action set is filtered and re-deductive write-back is performed; at the same time, the freeze reason reference and the release basis reference are written into the version chain.

[0020] The technical effects and advantages of the digital twin-based simulation method for power transmission and transformation engineering construction are as follows:

[0021] This invention transforms the uncertainty of underground objects from "data discrepancies" into "traceable corridor versions," and further transforms construction actions from "experience-based approval" into "release decisions constrained by corridor versions." Through the constraints of unified identification at the evidence level and spatial reference benchmarks, corridor boundaries and construction actions are aligned under the same coordinate semantics. The exposure results obtained from trial excavations can directly drive the convergence of credible corridors within a local range and form new versions. At the same time, the sources of discrepancies are decomposed into interpretable conflict types and the influence domain is limited, avoiding the boundless diffusion of uncertain information in subsequent deductions and construction organization.

[0022] Building upon this foundation, the extrapolation under the multi-credible corridor assumption elevates the question of "whether construction is feasible" from a single conclusion to a stability issue. Verification points are no longer selected based on experience but are prioritized to cover sensitive sections that could cause a reversal of the extrapolation conclusion and where verification efforts are more likely to lead to corridor convergence. During the action release phase, freezing and unfreezing are bound to the completion of verification and the effective implementation of the corridor version, enabling construction to continue without touching uncertain corridors. After the corridor is updated, affected actions are automatically reviewed and the complete decision-making chain is archived, thereby reducing the probability of accidentally touching underground pipelines and improving the reproducibility and accountability of the construction process in scenarios such as foundation pit excavation and support layout. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating a construction simulation method for power transmission and transformation projects based on digital twins, according to the present invention. Detailed Implementation

[0024] 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.

[0025] Example 1: Figure 1 This invention presents a simulation method for power transmission and transformation engineering construction based on digital twins, comprising:

[0026] S1: Construct evidence layers according to source, establish unified identifiers for underground objects and bind spatial reference benchmarks, and generate a credible corridor range volume containing source labels; construct the construction action sequence into an action envelope and establish a reference mapping;

[0027] S2: Map the exposure results of the verification points to a unified identifier, constrain the trusted corridor range to perform local alignment updates to form a new version of the corridor; merge the inconsistencies in the evidence layer into four types of conflicts and label the influence domain;

[0028] S3: Derive the constraint triggering chain on the set of credible corridor assumptions and output the feasibility label, construct the stability index and the verification value index and input them into the verification priority coefficient generator, output the minimum set of exploration closed loop points and complete the action set splitting;

[0029] S4: Read the gating result and verification priority coefficient, freeze the set of uncertain corridor actions and use verification completion and new corridor version effectiveness as release conditions, allow access without touching the alternative action set and archive the twin version chain.

[0030] This invention addresses the real-world contradiction of discrepancies and uncertainties in underground object data during power transmission and transformation construction. It transforms digital twins from a "single static model" into a "versioned model that converges with verification." The method first breaks down design drawings, as-built maintenance data, geophysical interpretation, and historical exposure records into evidence layers based on their sources. A unified identifier is established for each underground object, and it is bound to the same spatial reference benchmark. A credible corridor range with source labels is used to represent the possible location intervals of underground objects. Simultaneously, construction procedures such as excavation, support layout, machinery operation, and vehicle movement are converted into action envelopes, and a reference mapping is established between these action envelopes and the unified identifier. This allows for direct comparison and deduction of uncertain intervals of underground objects and construction-occupied areas under the same coordinate semantics.

[0031] When on-site test pits, excavations, and small trenches produce exposure results, the method maps these results to a unified identifier for underground objects, driving a local alignment update of the credible corridor volume to form a new version of the corridor. Simultaneously, it decomposes evidence inconsistencies into four categories: coordinate reference deviation, detection interpretation deviation, missing data, and path changes, labeling the affected spatial domain and the affected action reference domain to ensure consistency between the update range, the scope of influence, and the source tracing. Subsequently, a set of multiple credible corridor hypotheses is constructed on the same action sequence, deduce the constraint trigger chain between the action envelope and the corridor volume, quantify the degree of reversal of conclusions under different hypotheses, and locate the sensitive segments requiring verification. Then, combining the corridor convergence benefits brought by verification, a verification priority order is generated, forming a minimum set of excavation closed-loop points and a verification order. Finally, the action sequence is split into a set of actions that reach uncertain corridors and a set of actions that do not reach alternative corridors.

[0032] During the action release phase, the method reads the stability gating results and verification priority, sets freeze conditions for actions that reach uncertain corridors, and binds the release conditions to two thresholds: the completion of the verification event and the activation of the new corridor version. It keeps the alternative action set as executable and records the execution window. After verification, it triggers the corridor version update and re-infers the gating state for the affected actions. Finally, it archives the verification points, corridor version, gating criteria, action execution trajectory, and inference output into the twin version chain, so as to make the decision-making process reproducible and traceable under the same construction scenario.

[0033] Power transmission and transformation construction involves a complex environment where underground cable trenches, grounding electrodes, pipe rows, and existing municipal pipelines coexist. Design drawings, as-built operation and maintenance data, geophysical exploration interpretations, and historical exposure records often differ in their descriptions of location, orientation, and burial depth. Without a unified object and a unified spatial semantics, underground objects in digital twins will be scattered into multiple sets of misaligned representations, and construction actions will be difficult to express using the same spatial reference. Simulation and deduction will be unable to stably locate risk trigger points and form traceable inputs.

[0034] S101 Evidence Layer Decomposition and Spatial Benchmark Unification.

[0035] The excavation area of ​​the foundation pit usually contains coordinates from design data, coordinates from completion and operation, and coordinates from exploration results. When the coordinate benchmarks are inconsistent, the geometric position of the underground object will show overall translation, overall rotation, or even scale differences. Directly entering the subsequent matching will misjudge the coordinate error as object difference.

[0036] The evidence layers are divided into four categories based on their source: design drawing evidence layer, as-built operation and maintenance evidence layer, exploration and interpretation evidence layer, and historical exposure evidence layer. The spatial reference benchmark is selected using the coordinates of the construction survey control network, which is jointly determined by the established control points around the foundation pit and the fixed benchmark points in the station area. Control point pairs are acquired following the rule of prioritizing corresponding points. Corresponding points include the center points of manhole covers, pipeline valve wells, structure corner points, and survey stake points. Control point pair matching is first performed by name and type, and then consistency is verified using the point descriptions in the field measurement records. Alignment operations employ a rigid body alignment algorithm based on singular value decomposition, allowing the participation of a uniform scale factor. The operation order is to first calculate the centroid of the control point set in the coordinates of each evidence layer. The location is determined by subtracting the corresponding centroid from the control point coordinates to achieve decentralization. Then, singular value decomposition is performed on the covariance matrix of the decentralized point pairs to obtain the rotation relationship. The translation vector is then calculated to align the rotated centroids with the spatial reference reference centroid. At the same time, when a scale difference is detected, a uniform scale factor is introduced to minimize the mean square distance between point pairs. Outlier removal adopts random consistency sampling, randomly selecting the smallest set of point pairs to estimate the transformation parameters, statistically analyzing the residuals of the remaining point pairs and iteratively updating them. Point pairs whose residuals exceed the preset tolerance are removed from the control point pair set. After normalization, the geometry of each underground object in the evidence layer is unified to the spatial reference reference, the geometric unit is unified to the length unit, the attribute fields retain the original enumeration values, and the original coordinate system identifier is saved synchronously for traceability.

[0037] Example: The construction surveying team sets up control points at the four corners of the foundation pit. The design drawings provide valve well numbers, and the maintenance records provide well cover numbers. On site, the center points of well covers with the same numbers are used as control point pairs. Alignment calculations are used to align the coordinates of the maintenance records to the coordinates of the construction control network. Then, the coordinates of the detection and interpretation results are aligned to the same coordinates of the construction control network, so that the positions of the same valve well in the four types of evidence layers can be directly superimposed and compared. After completion, the geometry of each evidence layer can be superimposed and displayed on the same base map and maintain traceable alignment parameters.

[0038] S102 Underground Object Unified Identification Establishment and Cross-Layer Matching.

[0039] For the same underground object, there are often missing fields, inconsistent naming, and different geometric segmentation methods in different evidence layers. Directly using field equality matching is prone to missed matches, and directly using geometric proximity matching is prone to erroneous merging. Therefore, it is necessary to standardize before matching.

[0040] Attribute field normalization follows a field hierarchy rule, consisting of strongly constrained and weakly constrained fields. Strongly constrained fields include usage category, installation method, and ownership unit, while weakly constrained fields include material, caliber, and burial depth description. The normalization process first performs dictionary mapping on usage category and installation method, mapping synonyms and abbreviations to unified enumeration values, and then performs alias merging on ownership unit. Missing field compensation uses a proximity inheritance rule: if a structure record connected to the object to be matched exists in the same evidence layer and contains a missing field, then the missing field is filled using the structure record field and marked as the inference source. Candidate pair generation first uses strong constraint field consistency filtering, then uses geometric neighborhood constraint filtering. Geometric neighborhood constraints use Hausdorff distance calculation. The Hausdorff distance calculation order is to discretize the centerlines of the two objects into a set of sampling points of equal arc length, and then calculate the first... The maximum value of the Euclidean distance from each point in the first set of sampling points to the nearest point in the second set of sampling points is taken. Simultaneously, the maximum value of the distance from the second set of sampling points to the nearest point in the first set of sampling points is also taken. The larger of these two distances is then used as the Hausdorff distance. The neighborhood condition is derived from the upper bound of the basic radius mapping table. The criterion is that the Hausdorff distance is less than the combined boundary of the upper bound of the basic radius of the corresponding evidence layer for both objects, and the directional consistency requirement is met. Directional consistency is achieved by comparing the tangential angles of the center lines on several sampling segments. The unified identifier generation uses a disjoint-set data structure to merge connected components. The merging order is as follows: first, establish equivalence relations for objects selected through candidate pairs; then, calculate connected components for the equivalence relations; assign a unified identifier to each connected component; and simultaneously retain the mapping table from evidence layer objects to unified identifiers and the reverse mapping table from unified identifiers to the list of evidence layer objects.

[0041] Example: Design drawings divide the pipeline into two segments for recording, maintenance records record the same pipeline as one segment, and detection and interpretation results present the pipeline centerline as a curved segment; attribute normalization unifies the purpose of the three types of records as water supply pipeline; geometric neighborhood screening finds three segments of centerline that are adjacent to each other and continuous in direction near the valve well, and obtains the same unified identifier after lookup and set merging; the mapping table records the list of related objects in the three types of evidence layers, which is convenient for locating the same unified identifier when writing back the exposure point later.

[0042] S103 Trusted Corridor Scope Structure and Source Tag Registration.

[0043] When there is a positional inconsistency between underground objects in the evidence layer, a single centerline expression cannot cover the uncertain interval. The subsequent action envelope intersection judgment will jump due to the centerline deviation. The range volume expression can incorporate the uncertain interval into the deduction in a calculable way.

[0044] The credible corridor scope is represented by a tubular scope. The construction sequence is as follows: first, select a centerline for each unified identifier in each evidence layer. The centerline selection follows the longest continuous centerline priority rule, while preserving the connection point with the structure. The foundation radius value is derived from a mapping table of evidence layer type and data quality level. This mapping table is established offline, based on the classification of the deviation distribution between similar evidence layers and exposure points in historical projects. The local convergence factor is activated when historical or recent exposure points exist. The calculation sequence is: first, project the exposure point onto the centerline to obtain the arc length position; then, define the convergence support interval centered on the arc length position. The boundary of the convergence support interval is determined by the location of the exposure point. The influence range of the structure and the construction disturbance range are jointly determined. Then, within the convergence support interval, the local convergence factor is set as a dimensionless function that monotonically increases from the center to the boundary, with a smaller value at the center and not zero, and one outside the interval, thus obtaining a radius function that varies along the centerline. The range volume is generated by centerline sweeping. During the sweeping process, circular cross-sections are constructed point by point at the centerline sampling points and connected along the centerline to form a tubular body. The cross-section radius is taken from the position corresponding to the radius function. Source label registration binds a source label to each evidence layer range volume, and the range volumes of each evidence layer are merged to form a unified identifier corresponding to the overall credible corridor range volume. At the same time, the layered range volumes are reserved for subsequent construction of the multi-credible corridor hypothesis.

[0045] The S104 construction action sequence is constructed into an action envelope and labeled with action tags.

[0046] In the scenario of foundation pit excavation, risk triggering often occurs during spatial sweeping processes such as robotic arm swinging, vehicle movement, and hoisting of support components. The motion envelope transforms the process trajectory into a computable spatial volume, which can establish a geometric reference between the action reach range and the trusted corridor range volume.

[0047] Action labels are mapped from the process type field in the construction organization document, with values ​​limited to four categories: mechanical operation, excavation and advancement, support arrangement, and vehicle movement. The action envelope is constructed using a swept union of geometric primitives. The selection of geometric primitives follows the principle of covering the equipment shape and working posture. For mechanical operation, the geometric primitive is the union of the machine body envelope and the boom envelope; for vehicle movement, the geometric primitive is the vehicle body outline envelope; and for support arrangement, the geometric primitive is the support component hoisting posture outline envelope. Path parameters are obtained from the route alignment in the construction organization document and the on-site layout lines. Working posture parameters are obtained from the allowable posture range in the equipment operation procedures and on-site indicators. The working surface direction determined by the command is recorded in the form of discrete state sequence without using continuous control quantities; the operation order of sweep union is to first discretize the path into several key pose points along the motion process, then place the geometric primitives at each pose point to form an instantaneous occupancy volume, and then take the union of all instantaneous occupancy volumes to obtain the initial motion envelope; in order to avoid envelope breakage caused by discrete gaps, voxel closure processing is adopted. The voxel closure processing order is to first rasterize the initial motion envelope into a three-dimensional voxel occupancy set, then perform voxel dilation to fill the gaps, and then perform voxel erosion to restore the outer boundary, so that the shape of the motion envelope is continuous and there are no unreasonable holes.

[0048] Example: An excavator moves along a travel path at the edge of a foundation pit and swings its boom to excavate at several work positions. The motion process is discretized into travel pose points and boom pose points. At each discrete pose point, the outer contour of the excavator's upper body and the outer contour of the boom are placed in the corresponding position and merged to form an instantaneous occupied volume. The union of all instantaneous occupied volumes is used to obtain the initial motion envelope. After the initial motion envelope is rasterized, gaps appear. Voxel expansion fills the gaps, and voxel erosion restores the boundaries to obtain the continuous motion envelope.

[0049] S105 establishes the reference mapping between the action envelope and the trusted corridor range volume.

[0050] Subsequent simulations require quickly identifying which actions affect which underground objects in uncertain corridors, and calculating the separation distance upon impact for gating and sensitive segment location. Therefore, a searchable reference mapping table needs to be created.

[0051] The reference mapping table contains two types of fields: reach indication and separation distance. Reach indication calculation uses geometric intersection determination. The determination order is as follows: first, axis-aligned bounding box trees are constructed for the action envelope and the trusted corridor volume respectively as coarse spatial indices; then, candidate pairs of bounding box intersections are finely screened. Fine screening uses a hybrid representation of triangular meshes and voxels. The action envelope is represented by a set of voxels, and the trusted corridor volume is represented by a tubular mesh. Fine screening determines whether the action envelope voxels and the tubular mesh have voxel center points falling inside the tubular mesh or whether mesh faces intersect with voxel cubes. If an intersection exists, the reach indication value is taken; otherwise, the non-reach value is taken. The distance calculation employs a nearest-distance search. The search order is as follows: first, extract the set of boundary voxel center points from the action envelope voxel set; then, establish a multi-dimensional binary spatial index tree on the reliable corridor range volume grid; for each boundary voxel center point, search for the nearest point distance on the grid surface and take the minimum value as the separation distance; the separation distance is a non-negative length dimension, and the reach indicator is binary; the reference mapping table establishes a key-value index based on the action identifier and the unified identifier. The index structure adopts a two-level hash table, with the first-level key being the action identifier and the second-level key being the unified identifier, and the values ​​being the reach indicator and the separation distance, thus satisfying the need for rapid access to the relationship between actions and underground objects in subsequent multi-hypothesis deductions.

[0052] By splitting the evidence layer according to its source and establishing a unified identifier and spatial reference benchmark for underground objects, the credible corridor range volume carries uncertain intervals by superimposing source labels on geometric boundaries. The construction action sequence is transformed into an action envelope with action labels and a reference mapping is established with the unified identifier of underground objects, so that underground objects and construction occupancy form a computable and traceable association structure under the same twin coordinate semantics.

[0053] On-site test pits, exploratory excavations, and small trenches are the most definitive verification methods in power transmission and transformation construction. However, the results are often recorded in point form or described in words. If they cannot be mapped to a unified identifier and drive the correction of corridor boundaries, the verification information will only remain at the record level and will not change the credible boundaries of the twin model. At the same time, if the differences between evidence layers are not classified and the influence domain is not limited, updates will introduce new inconsistencies and destroy version traceability.

[0054] S201 Verification Event Registration and Exposure Results Structure.

[0055] The exposure results obtained from the excavation of the foundation pit are usually obtained from three types of methods: manual measurement, total station measurement, and point cloud measurement. When the recording methods are not uniform, the exposure results cannot be stably written back to a unified identifier and drive the evolution of the trusted corridor range version.

[0056] The verification event registration includes five types of fields: verification point location, exposed point set, exposed description set, measurement method identifier, and measurement resolution scale. Verification point locations are expressed using spatial reference coordinates and the coordinate units are stored. Exposed point sets are expressed using the same spatial reference coordinates and are required to cover continuous traverse segments of the exposed object. The measurement method identifier is limited to one of three types: total station measurement, manual ruler measurement with stakeout verification, or point cloud measurement. The measurement resolution scale is limited to a length dimension greater than zero and is derived from the minimum instrument reading or point cloud sampling interval corresponding to the measurement method identifier. A one-to-one correspondence is formed between the measurement resolution scale and the measurement method identifier to avoid a lack of basis for subsequent radius convergence lower bounds.

[0057] Example: A section of water supply pipeline is exposed in a test pit. The construction worker uses a total station to measure multiple points on the center line of the top of the pipe and records them as a set of exposed points. The verification point is taken as the center position of the test pit. The measurement method is identified as total station measurement. The measurement resolution scale is recorded as the length scale corresponding to the minimum reading of the total station. The exposure description set is recorded as the connection information between the water supply pipeline and the valve well.

[0058] S202 verification points are mapped to a unified identifier and the target object is locked.

[0059] There may be overlapping trusted corridor ranges of multiple underground objects near the same verification point. Simply mapping based on geometric proximity can easily write the exposure results into the wrong unified identifier, causing subsequent local alignment updates to converge to the wrong corridor boundary.

[0060] The mapping process prioritizes point inclusion retrieval, followed by nearest distance retrieval, and finally exposure description consistency verification. Point inclusion retrieval is based on the layered corridor range volume and centerline radius expression output in step S1. The determination order is as follows: search for the nearest projected position of the verification point along the layered centerline of each candidate unified identifier; calculate the straight-line distance from the verification point to the centerline projection position; and read the radius function value corresponding to the projection position. If the straight-line distance does not exceed the radius value, the verification point is determined to fall within the corresponding layered corridor range volume. If any layered corridor range volume meets the inclusion determination, the unified identifier is added to the candidate set. Nearest distance retrieval is used when the candidate set is empty. The calculation order is as follows: search for the nearest projected position of the verification point on the layered centerline of the unified identifier and calculate the straight-line distance; then subtract the radius function value to obtain the distance to the corridor boundary; select the unified identifier with the smallest boundary distance as the candidate mapping result. The exposure description consistency check uses the evidence layer to unified identifier mapping table output in step S1 to read the purpose category and laying method fields, and compares them with the purpose category and laying method in the exposure description set. If inconsistencies are found, inconsistent unified identifiers are removed from the candidate set. If multiple unified identifiers still exist, the target object is locked according to the principle of minimum boundary distance. After the mapping is completed, a binding record from the verification event to the unified identifier is formed and written into the version chain entry, so that the unified identifier can be directly referenced in subsequent local alignment updates.

[0061] The S203 local alignment update rule is executed and a new version of the corridor is generated.

[0062] The exposure results reflect the true location of underground objects in local space. The credible corridor range needs to converge within the exposure neighborhood and maintain the continuity of the external segments. Directly replacing the centerline globally would destroy the traceability of the evidence layer and the spatial continuity.

[0063] The input for the local alignment update rule includes a unified identifier, verification points, exposed point set, measurement resolution scale, layer centerline, layer radius function, and layer corridor range. The centerline projection point is obtained using an arc-length sampling search. The search order is to sample the layer centerline at equal arc lengths to form a sequence of sampling points, then calculate the straight-line distance from the verification point to each sampling point, and take the minimum value corresponding to the sampling point as the projection point. The residual vector is obtained by subtracting the projection point coordinates from the verification point coordinates. The residual vector direction points to the spatial direction that needs correction, and the residual vector length is a length dimension. The support interval is determined jointly by the exposed point set coverage length and the excavation influence range. The determination rule is to take the arc length position corresponding to the projection point as the center, extend forward and backward to the exposed point set's coverage endpoint, and then extend outward by a construction disturbance buffer section. The length of the buffer section is recorded as a construction parameter in the verification event entry.

[0064] The local relocation of the centerline uses an allocation function to distribute the residual vector along the arc length. The allocation function is a piecewise function of triangles, with the maximum value taken at the center of the support interval, zero at the boundary of the support interval, and zero outside the support interval. The relocation order is as follows: calculate the distance from the arc length position of each sampling point in the support interval to the center of the support interval, normalize the distance to the interval between zero and one, multiply the value of the allocation function by the residual vector to obtain the displacement vector of the sampling point, add the displacement vector to the coordinates of the sampling point to form the updated centerline sampling point sequence, and then reconstruct the layered centerline from the updated sampling point sequence.

[0065] The local convergence of the radius follows the non-expansion constraint and the lower bound constraint of the resolution scale. The convergence order is as follows: calculate the normalized distance from each arc length sampling point in the support interval to the center of the support interval, and then generate a monotonically increasing convergence factor based on the normalized distance. The convergence factor at the center of the support interval is minimized and kept greater than zero, the convergence factor at the boundary of the support interval is set to one, and the convergence factor outside the support interval is set to one. Then, the smaller value of the original radius value and the convergence factor is taken to form the non-expansion constraint radius. Finally, the larger value of the non-expansion constraint radius and the measurement resolution scale is taken to form the lower bound constraint radius, and the updated radius function is obtained.

[0066] When multiple exposure points simultaneously affect the same unified identifier, the update order is executed according to the verification event timestamps. If multiple exposure points exist at the same timestamp, the geometric median vector is calculated first for the residual vector set, and then a centerline relocation is performed. The geometric median vector is defined as the target vector that minimizes the sum of straight-line distances from the residual vector set to the target vector. After the centerline and radius are updated, the layered corridor volume is re-sweeped and generated according to the updated centerline and updated radius functions. The layered corridor volume is then used to form the new version of the corridor corresponding to the unified identifier, while the old version is retained as a historical branch and the version index increment relationship is recorded.

[0067] Example: An exposed section of cable tunnel was excavated. The exposed point set showed that the tunnel's orientation slightly shifted near the valve well. The residual vector was obtained after the verification points were projected onto the centerline of the completed operation and maintenance evidence layer. The support interval covered the exposed sections on both sides of the valve well. The centerline sampling points were translated point by point within the support interval according to the triangular distribution function. The radius function shrank at the center of the support interval according to the convergence factor and was constrained by the measurement resolution scale. After the update, the corridor boundary fit the exposed section and maintained continuity outside the support interval.

[0068] S204 Conflict Decomposition and Labeling of Affected Spatial Domains and Affected Action Reference Domains.

[0069] If inconsistencies in the evidence are not categorized, coordinate baseline errors, detection and interpretation errors, missing data, and path changes will be confused into one type of problem, making it impossible to specifically address the affected sections and actions in subsequent deductions.

[0070] Conflict decomposition is performed around the hierarchical centerline associated with the unified identifier and the updated set of residual vectors. Coordinate baseline deviation is determined using a joint approach of direction consistency and length consistency. Direction consistency is calculated by taking the cosine of the angle between any two residual vectors, using the vector dot product divided by the vector length product. If the cosine is above a preset lower limit, the direction is considered consistent. Length consistency is calculated by sorting the residual vector lengths in ascending order and calculating the maximum difference between adjacent lengths. If the maximum adjacent difference does not exceed a preset upper limit of discreteness, the length is considered consistent. When both direction and length are consistent, the conflict is classified as coordinate baseline deviation.

[0071] The detection interpretation bias determination targets the detection interpretation evidence layer. The calculation order is as follows: within the support interval, calculate the turning sequence formed by three consecutive points of the centerline sampling point sequence; then, calculate the turning sequence of the trend line segment sequence obtained by fitting the exposed point set. If the two types of turning sequences show a continuously opposite turning trend within the support interval and the endpoint connecting structures are consistent, then the conflict is classified as a detection interpretation bias. The data missing determination directly reads the evidence layer from the unified identifier mapping table. If a mapping entry is missing for a certain evidence layer, the conflict is classified as data missing.

[0072] Path change determination uses topological clues from the exposed point set. The calculation sequence is as follows: perform piecewise straight-line fitting on the exposed point set and find inflection points; then compare the number of inflection points and the connection relationships of the corresponding structures with the centerlines of each evidence layer. If the exposed point set has new inflection points or new branches and the structure connection set has changed, the conflict is classified as a path change. The affected spatial domain is defined as the updated local volume of the corridor corresponding to the support interval. The generation sequence is as follows: take the local segment of the updated layered corridor range volume corresponding to the arc length of the support interval and merge the layered local segments to form the affected spatial domain.

[0073] The affected action reference domains are obtained by filtering the reference mapping table output in step S1. The filtering order is to first perform a coarse screening using the bounding box of the action envelope and the bounding box of the affected spatial domain, and then perform a fine screening intersection determination on the coarse screening candidates. The fine screening determination uses the intersection test of voxels and tubular meshes to obtain the set of actions that intersect with the affected spatial domain and register them as affected action reference domains.

[0074] S205 version chain registration and rollback rules.

[0075] The same unified identifier will form multiple local convergence and multi-branch versions under the action of different verification events. The lack of version chain registration will cause subsequent inference inputs to be unreproducible and make it impossible to explain the basis for action freezing and release.

[0076] The version chain entry registration content includes a unified identifier, verification event number, verification point location, exposure point set index, measurement method identifier, measurement resolution scale, support interval description, corridor version index before update, corridor version index after update, conflict classification, affected spatial domain index, affected action reference domain index, and timestamp. Rollback triggering conditions are limited to verification event cancellation records or subsequent verification events forming opposite residuals for the same support interval, and the conflict classification changing from coordinate baseline deviation to path change. The rollback execution order is to restore the current corridor version pointer to the rollback target version and retain the rolled-back version as a historical branch. The historical branch retains its source and triggering reason to avoid version chain breaks and ensure that simulation and reproduction experiments can be replayed according to the version index.

[0077] The verification event binds the exposure results with the unified identifier of the underground object and triggers a local alignment update of the trusted corridor range. The corridor boundary forms a new version through local relocation and range convergence. Inconsistencies in the evidence layer are categorized into coordinate reference deviation, detection interpretation deviation, data loss and path change, and the affected spatial domain and affected action reference domain are marked, so that the corridor version evolution, conflict attribution and impact propagation are consistent.

[0078] When there are still uncertain intervals at the corridor boundaries, the simulation conclusions under a single corridor version are often sensitive to the selection of evidence. This manifests as the same action appearing as feasible or infeasible under different assumptions, making it difficult for construction organizations to arrange verification points and the order of action releases accordingly. It is necessary to explicitly expand the differences in evidence and link the verification input and corridor convergence benefits to calculable indicators.

[0079] Step S3 selects the feasible flip ratio and the corridor shrinkage benefit density separately to participate in the comprehensive analysis to generate the verification priority coefficient. The core is that the two respectively characterize the disturbance intensity of the uncertain corridor on the construction deduction conclusion and the ability of verification input to inhibit corridor convergence. Moreover, the two have the same source and do not substitute for each other: the feasible flip ratio comes directly from the feasibility label flip statistics of the same action sequence under the multiple credible corridor assumption. It reflects that once the evidence divergence enters the deduction chain, it will push the action judgment to an unstable state. It can locate the "sensitive segment that is easily disturbed by evidence differences" from the action sequence. The corridor shrinkage benefit density comes from the ratio of the expected range volume reduction of the local alignment update operator to the volume represented by the verification workload. It reflects the efficiency of one verification for the convergence of the uncertain interval. It can distinguish the "verification points that can better eliminate the uncertainty boundary under the same input" from the candidate set. Combining the two to generate a verification priority coefficient avoids the imbalance in verification resource allocation caused by relying solely on sensitivity gating, and also avoids the omission of highly sensitive flipping segments due to relying solely on convergence efficiency ranking. At the same time, it reduces the dependence on multi-parameter thresholds and empirical rules, ensuring that gating and ranking maintain consistent judgment criteria and traceable interpretation chains when evidence changes, corridor version updates, and construction actions are adjusted. This allows for more stable control of the release rhythm of actions reaching uncertain corridors and reduces the risk of erroneous releases in the scenario of foundation pit excavation and support layout.

[0080] S301 Multiple Credible Corridor Hypothesis Set Construction.

[0081] The Trusted Corridor Range Library stores the hierarchical corridor ranges and source tags. The corridor version chain stores the new version of the corridor after local alignment and updates. If only the overall union corridor is selected in the construction simulation, the differences in the evidence layers will be masked. If only a single-layer corridor is selected in the construction simulation, the uncertain intervals will be missed. Therefore, the multi-trusted corridor hypothesis set is used to explicitly expand the evidence divergence and maintain the traceability of the version chain.

[0082] For each unified identifier, the new version of the effective corridor is read from the corridor version chain, and the set of layered corridor ranges is read. A set of credible corridor hypotheses is generated by combining source tags. The combination rules include three types: single-layer hypothesis, two-layer joint hypothesis, and full-layer joint hypothesis. Boundary closure operation is performed on the boundary of each hypothetical corridor. The boundary closure operation adopts voxel rasterization, voxel dilation, and voxel erosion. The voxel resolution scale is the smaller of the measurement resolution scale registered in the corridor version chain and the minimum spacing of the construction layout line, while maintaining the length dimension. When the conflict type is coordinate reference deviation, a translational disturbance branch is added to the hypothesis set. The translational disturbance branch applies translational compensation in the residual direction along the center line segment corresponding to the affected spatial domain and re-sweeps to obtain the compensated corridor. When the conflict type is path change, a topology branch is added to the hypothesis set. The topology branch fits the inflection point along the exposed point set and replaces the center line segment at the inflection point. Then, it is swept with the same radius function to obtain the changed corridor. The hypothesis set output registers the source tag set and version index at the same time to avoid ambiguity in subsequent inference inputs.

[0083] S302 Constraint Trigger Chain Deduction and Feasibility Tag Generation.

[0084] The action envelope library transforms mechanical operations, excavation and advancement, support layout, and vehicle movement into geometric occupancy volumes. After introducing a set of credible corridor assumptions, the same action may have differences in reach and clearance under different assumed corridors. The constraint trigger chain is used to fix the triggering cause to the action index and unified identifier, avoiding the inability to locate the triggering source by only outputting feasible or infeasible conclusions.

[0085] The action envelope and action label are read sequentially according to the construction action sequence. For each action envelope and each credible corridor hypothesis corridor, reach determination and clearance determination are performed. The reach determination uses bounding box tree coarse screening and then performs voxel and tubular grid fine screening intersection test. The clearance determination uses the minimum Euclidean distance from the set of voxel center points of the action envelope boundary to the nearest point of the corridor boundary grid. The separation distance value ranges from zero to positive infinity and maintains the length dimension. The clearance trigger criterion is given by the clearance threshold registered in the construction organization document. The clearance threshold value ranges from zero to positive infinity and maintains the length dimension. When the separation distance is less than the clearance threshold, it is recorded as clearance trigger. The trigger chain records the action index, unified identifier, trigger type and separation distance of the first trigger according to the action sequence. At the same time, a feasibility label is output for each hypothesis. The feasibility label value is limited to feasible and infeasible. The two types of values ​​are determined by whether the trigger chain has a reach trigger or a clearance trigger.

[0086] Example: The support arrangement action includes hoisting the steel support and turning it into place. The action envelope covers the boom slewing sweep area. For a certain credible corridor, it is assumed that the corridor is close to the edge of the foundation pit. The intersection test shows that the boom slewing area and the corridor boundary intersect. The trigger chain records the support arrangement action index and the corresponding unified identifier and marks the trigger. For another credible corridor, it is assumed that the corridor does not intersect after it expands outward, but the separation distance is lower than the net distance threshold. The trigger chain records the same action index but marks the net distance trigger.

[0087] S303 Feasible flip ratio calculation and sensitive segment location.

[0088] When the same action sequence yields inconsistent conclusions of feasibility and infeasibility under different credible corridor assumptions, the construction decision may exhibit a conclusion reversal phenomenon. The feasibility reversal ratio is used to quantify the reversal frequency and fix the concentrated reversal section as the sensitive section. The sensitive section is used to limit the spatial range of the generation of verification candidate points.

[0089] For each unified identifier, a baseline hypothesis is selected from the set of credible corridor hypotheses. The selection of the baseline hypothesis follows a parallel rule of prioritizing historical exposure evidence and the latest version index. When there is a conflict in the parallel rules, the source label confidence sequence is used to fix the selection order. The feasible flip ratio is calculated by comparing the feasibility labels of other hypotheses with the feasibility labels of the baseline hypothesis one by one, counting the number of inconsistencies, and then using the ratio of the number of inconsistencies to the number of comparisons to generate the feasible flip ratio. The feasible flip ratio ranges from zero to one and is dimensionless. Sensitive segment location starts from the first trigger anchor point of the trigger chain. The anchor point is determined by the nearest point of the action envelope boundary of the triggering action. The anchor point is then projected along the construction layout line to generate a mileage expression. The arc length interval of the corridor centerline corresponding to the triggering unified identifier in the affected spatial domain is mapped to the same mileage expression to form a sensitive segment interval. The sensitive segment interval is expressed by a combination of the start and end mileage plus the associated structure identifier and written into the sensitive segment set.

[0090] S304 Corridor Revenue Density Estimation and Validation Candidate Site Set Generation.

[0091] Selecting verification points within sensitive sections requires consideration of both corridor convergence magnitude and on-site workload. Step S2 has defined local alignment update rules and formed a corridor version chain. Step S3 uses the local alignment update rules to estimate the corridor convergence magnitude brought about by verification, thereby providing an interpretable geometric basis for the ranking of verification points.

[0092] A set of candidate verification points is generated within the sensitive segment set. The candidate point generation employs the minimum net distance anchoring rule, which selects the nearest point on the action envelope boundary corresponding to the minimum separation distance from the separation distances recorded in the trigger chain under the baseline assumption. This nearest point is then projected onto the center of the normal section of the baseline assumption centerline to form the candidate point. Simultaneously, the candidate point inherits the support interval length and measurement method identifier to satisfy the input constraints of the local alignment update rule. The expected range reduction estimation order is as follows: candidate points are written as hypothesis verification points into temporary verification events; the local alignment update rule is called to perform only radius convergence while keeping the centerline unchanged; and then within the support interval... For discrete cross-sections of equal arc length, the difference between the area of ​​the cross-section before and after the update is calculated for each cross-section. Then, the difference in cross-sectional area is discretized and summed along the support interval to obtain the expected range volume reduction. The expected range volume reduction ranges from zero to positive infinity while maintaining the volume dimension. The volume representing the verification workload is directly given by the planning geometry of the test pit or small slot, and its value ranges from greater than zero to positive infinity while maintaining the volume dimension. The corridor shrinkage benefit density is the ratio of the expected range volume reduction to the volume representing the verification workload. The corridor shrinkage benefit density is dimensionless and its value is greater than zero. The candidate point set is output while simultaneously registering the support interval expression and the sensitive segment interval expression to avoid the inability to determine the coverage relationship of candidate points.

[0093] Example: The excavation advance action triggers a clearance near the edge of the foundation pit. The minimum clearance anchor point falls on the boundary of the bucket rotation sweep. The anchor point is projected onto the normal section of the corridor centerline to obtain the candidate point. After the candidate point is slotted on site, the cable duct is exposed. The local alignment update rule converges the corridor radius within the support interval. The corridor boundary moves closer to the centerline. The sensitive section is shrunk from the structure on both sides of the duct to a shorter interval on the layout line.

[0094] S305 Verification Priority Coefficient Generation and Action Set Splitting.

[0095] S3051 stability gating and candidate set determination.

[0096] The feasible flip ratio corresponding to each unified identifier is read and compared with the stability gating threshold. Unified identifiers that meet the threshold condition are added to the verification trigger set. Only the verification candidate points associated with the verification trigger set are retained to form a candidate set to be sorted. This process prioritizes sensitive segments with unstable inference conclusions in the verification process, avoiding the dispersion of verification resources to low-disturbance segments.

[0097] S3052 Training Sample Construction and Sorting Label Generation.

[0098] Historical verification events are extracted from the corridor version chain, and sample records are formed according to the verification points. The sample features are fixed as the corridor shrinkage benefit density and the feasible flip ratio of the unified identifier. The sample labels are sorting labels, not classification labels. The sorting labels are generated based on the relative priority relationship between the actual corridor version convergence amplitude and the actual verification workload characterization volume of the historical events.

[0099] To ensure sample consistency, records with incomplete version rollbacks, missing verification records, and missing action trajectories were removed.

[0100] The following are examples of data; the values ​​are for illustrative purposes only and are used to illustrate the sample organization method:

[0101] Table 1. Training Sample Example Data Table

[0102]

[0103] S3053 Monotonic Constraint Decision List Model Construction.

[0104] A monotonic constraint decision list model is used as the validation priority coefficient generator. The model input consists of two fixed features: the feasible flip ratio and the corridor shrinkage benefit density; the model output is the validation priority coefficient, which is limited to the range of zero to one and is dimensionless.

[0105] The model rule terms adopt a "condition interval + score" structure. The condition interval is obtained by binning and combining two input features, and the score is the score assigned to the rule terms. The rule terms are arranged into a decision list in priority order. If no rule term is matched, it falls into the default rule term.

[0106] Monotonic constraints are applied in two directions: the rule score does not decrease when the feasible flip ratio increases, and the rule score does not decrease when the corridor benefit density increases; if a rule term is violated during the training phase, monotonic repair is performed and the rule score is written back to ensure that the model output is consistent with engineering understanding.

[0107] Example of S3054 model optimization, iteration and parameter setting.

[0108] The training objective is to maximize pairwise sorting consistency, combined with a complexity penalty and a monotonic violation penalty. The optimization process employs a closed-loop workflow of "candidate rule generation – rule selection – score update – monotonic correction – resampling iteration".

[0109] The first step is to generate candidate rules, which are derived from binning combinations of two features;

[0110] The second step is to screen out weak rules based on the minimum number of samples covered.

[0111] The third step is to update the rule order and rule score based on the consistency improvement amount.

[0112] The fourth step is to perform monotonic repair to eliminate the breach of contract;

[0113] The fifth step involves resampling missorted samples and samples that are difficult to execute, and then proceeding to the next iteration.

[0114] Example parameter settings are as follows: Feasible flipping score bin number is 6, corridor benefit density score bin number is 6, rule item limit is 10, minimum coverage sample size for a single rule is 20, complexity penalty coefficient is 0.02, monotonic violation penalty coefficient is 1.5, maximum iteration rounds are 120, and early stop rounds are 15. These parameters were determined through cross-validation of historical projects and can be recalibrated according to regional construction organization differences.

[0115] S3055 Candidate Point Ranking and Minimum Exploration Closed-Loop Point Set Generation.

[0116] The minimum set of exploration closed-loop points is generated under the coverage greedy rule. The coverage greedy rule selects candidate points from high to low according to the verification priority coefficient, and updates the set of sensitive segments that have not yet been covered after each selection. The coverage determination is based on the intersection of the support interval of the candidate point and the interval of the sensitive segment, until the entire set of sensitive segments is covered. The coverage greedy rule outputs the minimum approximate solution that satisfies the full coverage constraint and registers the selection order.

[0117] Input the candidate set to be sorted into the trained monotonic constraint decision list model to obtain the verification priority coefficient of each candidate point and generate a candidate point sorting sequence in descending order.

[0118] The minimum exploration closed-loop point set is generated using a coverage greedy rule: candidate points are selected point by point according to the sorting sequence, and the set of sensitive segments that have not yet been covered is updated after each selection. The coverage determination is based on the intersection relationship between the support interval of the candidate point and the interval of the sensitive segment. When the set of sensitive segments is completely covered, the selection stops, and the minimum exploration closed-loop point set and the verification order are output.

[0119] S3056 Action Set Splitting and Step S4 Interface Output.

[0120] The action set is split based on the reference mapping table and the verification trigger set. Actions with a reach indication that are reached and associated with a unified identifier belonging to the verification trigger set are written into the reach uncertainty corridor action set; the remaining actions are written into the non-reach alternative action set in the original action order.

[0121] At the same time, the action is registered to the unified identifier of the reach source list, and together with the verification priority coefficient, candidate point sorting sequence, and minimum exploration closed loop point set, it is output to step S4 for binding the freeze condition and release condition.

[0122] The multi-evidence layer and the multi-credible corridor hypothesis set drive the constraint trigger chain deduction and output feasibility labels on the same action sequence. The feasibility flip ratio characterizes the stability of the conclusion and locates the sensitive section. The narrowing corridor benefit density characterizes the efficiency of the verification input on the corridor convergence. The two enter the verification priority coefficient generator of monotonic constraints to form the minimum exploration closed loop point set and verification order. At the same time, the action sequence is split into the action set of reaching the uncertain corridor and the action set of not reaching the alternative.

[0123] During the action release phase, it is necessary to avoid the risk of accidental triggering due to reaching uncertain corridors, while maintaining the construction continuity of non-reachable sections. If there is no clear threshold for freezing and unfreezing, on-site execution is prone to organizational deviations such as "reaching before verification is completed" or "repeated freezing even after the update has taken effect". At the same time, if the affected actions are not reviewed after the corridor version is updated, the simulation results and the on-site trajectory will deviate from each other in a way that cannot be reproduced.

[0124] S401 Action Release Input Aggregation and Release Unit Generation.

[0125] The core challenge in the action release phase is that reaching the uncertain corridor action set requires waiting for the verification of the closed loop to be completed, while not reaching the alternative action set requires maintaining the continuity of construction organization. A unified release unit is used to bind actions, unified identifiers, corridor version chain indexes, and minimum exploration closed loop point set indexes to the same record caliber, avoiding the lack of traceable references when freezing and unfreezing.

[0126] From step S3, read the stability gating results, verification priority coefficients, minimum exploration closed-loop point set, uncertain corridor reach action set, and non-reach alternative action set, and read the reach source list of actions to unified identifiers from the reference mapping table. Generate release unit records for each action sequence. The release unit record contains action identifier, action tag, reach source unified identifier set, associated corridor version chain effective version index, and associated minimum exploration closed-loop point set index. The associated minimum exploration closed-loop point set index is generated according to the support interval intersection rule, which is that the sensitive segment interval corresponding to the unified identifier of the reach source intersects with the support interval of the verification point. The reach source unified identifier set is limited to a subset of the unified identifier set, and the associated corridor version chain effective version index is limited to a valid version index existing in the corridor version chain. The release unit record is written to the version chain and used as input for freeze / unfreeze calculation.

[0127] S402 Freeze condition generation and unbind condition binding.

[0128] The risk of reaching an uncertain set of corridor actions does not come from the actions themselves, but from the corridor boundary not converging due to incomplete verification events or the new version of the corridor not being effective. Freezing and unfreezing conditions need to fall on inspectable state fields to avoid relying on empirical standards.

[0129] For each verification point in the minimum exploration closed-loop point set, register the verification event completion status. The verification event completion status is limited to two states: completed and incomplete. The criteria for determining the completion status are that the verification event has been registered, the exposure result has been mapped to the unified identifier, and the verification record entry has been written into the corridor version chain. For each unified identifier, register the corridor new version effectiveness status. The corridor new version effectiveness status is limited to two states: effective and ineffective. The criteria for determining the effectiveness status are that the corridor new version has been written into the corridor version chain and the effective version index has been switched to the corridor new version index. For each action in the uncertain corridor action set, read the associated verification point set index and the unified identifier set of the reach source, and perform the de-judgment process:

[0130] The first step is to check the completion status of the verification events of all verification points pointed to by the associated verification point set index. If any verification point is in an incomplete state, the release decision is set to not released.

[0131] The second step is to check the new version of the corridor for all unified identifiers in the unified identifier set of the source. If any unified identifier is in an ineffective state, the cancellation decision is set to not cancelled.

[0132] The third step is to set the release judgment to released if no incomplete or ineffective state is found in the first and second steps. The freeze judgment is obtained by inverting the release judgment. If the release judgment is not released, the action status is set to frozen and the freeze reason reference is written. The freeze reason reference points to the associated verification point set index and the unified identifier set of the reach source. If the release judgment is released, the action status is set to released and the release basis reference is written.

[0133] Example: The excavation and advancement action reaches the unified identification of the cable bar and is associated with two verification points. After the first verification point is completed and the exposure result is recorded on site, the second verification point is still in an incomplete state. The release judgment remains incomplete, and the action status remains frozen. On site, the vehicle movement is released according to the non-reach alternative action set and passes along the designated channel. After the second verification point is completed and the new version of the corridor is triggered, the release judgment is switched to released, the excavation and advancement action is released, and the release basis reference is recorded.

[0134] S403 releases the action execution window and archives the execution trajectory.

[0135] The execution process of not touching the set of alternative actions will also change the construction space occupation. If the execution window and execution trajectory archive are missing, the experimental input cannot be reproduced after the new version of the corridor takes effect, resulting in the deduction output being disconnected from the on-site trajectory.

[0136] For each action in the set of non-reachable alternative actions, the action status is set to "allowed" while maintaining the action order. The execution window is expressed using a start timestamp and an end timestamp. The start timestamp is the recorded time when the device enters the action envelope and the end timestamp is the recorded time when the device leaves the action envelope and the end timestamp. The execution window satisfies the condition that the start timestamp is earlier than the end timestamp. The action execution trajectory is expressed as a pose sequence, which is obtained by aligning the device positioning record and the construction record. The alignment uses timestamp alignment and expresses the pose using a spatial reference. When missing segments of the pose sequence are filled in by interpolation, the interpolation segment needs to be marked in the trajectory metadata. The pose sequence is used for consistency verification with the geometric envelope of the action envelope. The consistency verification criteria are that the entire trajectory point set falls within the occupied area of ​​the sweep volume of the action envelope and allows the boundary margin corresponding to the measurement resolution scale. For actions in the set of actions that have been released from the uncertain corridor, the execution window and execution trajectory are also executed and archived. For frozen actions, the freeze start timestamp is registered and the freeze reason reference is written into the version chain entry.

[0137] S404 verification complete, triggering the new version of the corridor to take effect and the affected actions to be re-inferred and rewritten.

[0138] After the verification event is completed, the new version of the corridor will take effect, which will change the reach determination and net distance determination. The freeze action needs to be re-analyzed with the new version of the corridor before it is lifted, and the release action also needs to form a traceable comparison record of the differences in the corridor boundary before and after it takes effect.

[0139] When the verification event completion status of any verification point changes from incomplete to complete, the corresponding unified identifier is read from the verification record entry, and a local alignment update rule is triggered to generate a new corridor version. The new corridor version is written to the corridor version chain, and the effective version index is switched. At the same time, the effective status of the new corridor version corresponding to the unified identifier is set to effective. The set of affected actions is obtained by filtering from the reference mapping table. The filtering criteria are that the unified identifier set of the action's reach source contains the unified identifier of the effective switch, and the action belongs to the set of actions that reach uncertain corridors or the set of alternative actions that do not reach. For each action in the set of affected actions, a re-deduction and write-back process is performed. The re-deduction and write-back process involves reading the action envelope and re-executing the intersection and nearest distance judgments under the new corridor version pointed to by the effective version index in the corridor version chain. A new constraint trigger chain record is generated and written to the version chain. At the same time, the old trigger chain record index is associated with the new trigger chain record index. For each action set that reaches an uncertain corridor, the release decision is re-executed and the freeze decision is updated. When the freeze decision is changed from freeze to release, the release timestamp is written and the freeze reason reference and release basis reference are retained. When a new constraint trigger chain record is triggered, the first trigger action identifier, the first trigger unified identifier, the trigger type and the trigger distance are written. The trigger distance retains the length dimension and the value is non-negative.

[0140] S405 version chain archive and reproduction summary identifier generation.

[0141] The evidence chain for action release and verification convergence involves the migration of the corridor version chain effective index, stability gating results, verification priority coefficient, freeze release status, execution window, execution trajectory, and trigger chain deduction output. The lack of unified archiving will lead to the inability to locate the complete input set when reproducing the same scenario.

[0142] The release unit records, freeze reason references, release basis references, corridor new version effective status change records, affected action set filtering records, re-deduction write-back records, execution window records, and execution trajectory records generated in this step are all uniformly written into the version chain archive entries, and fixed sorting rules are registered in the archive entries. The reproduction summary identifier is generated using a summary algorithm. The summary algorithm input consists of four types of sequences serialized and encoded in a fixed order and then concatenated. These four types of sequences include the corridor version chain effective version index sequence, action identifier sequence, minimum exploration closed-loop point set sequence, and constraint trigger chain record index sequence. The serialization encoding rules include fixed field separators, coordinate expression precision consistent with measurement resolution scale, action identifier sorting according to action sequence order, point sorting according to verification priority coefficient, and trigger chain index sorting according to action sequence order. The concatenated string is input into the summary algorithm to obtain a fixed-length summary identifier. The fixed-length summary identifier is written into the version chain archive entries and used as a reproduction retrieval key to locate all related record sets in the same construction scenario.

[0143] The action release process reads the stability gating results and verification priority coefficients, sets freeze conditions for actions that reach uncertain corridors, and binds the release conditions to the dual thresholds of verification event completion and corridor version update completion. It keeps the action release process open and records the execution window if it does not reach alternative action sets. After verification is completed and the new corridor version takes effect, it re-deduces the feasibility of the affected actions and writes back the gating status. The verification points, corridor versions, gating criteria, action execution trajectories, and deduction outputs are archived together in the twin version chain to support reproduction and traceability.

[0144] Specifically, the above are merely preferred embodiments of this application and are not intended to limit this application.

[0145] The thresholds or preset parameters can be pre-calibrated through offline simulation testing or set to fixed values ​​according to on-site operating procedures.

[0146] In the description of this specification, references to terms such as "an embodiment," "example," and "specific example" indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0147] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A simulation method for power transmission and transformation engineering construction based on digital twins, characterized in that, Including the following steps: S1: The design drawing evidence layer, the completion and operation and maintenance evidence layer, the exploration and interpretation evidence layer, and the historical exposure evidence layer are combined into an evidence layer set. A unified identifier is established for the underground object and a spatial reference benchmark is bound to it. Based on the unified identifier, the center line is extracted from each evidence layer, and a tubular range volume is constructed to express the uncertain range of the underground object's location as a credible corridor range volume. A source label is registered for the credible corridor range volume. The construction action sequence is constructed into an action envelope and a reference mapping between the action envelope and the unified identifier is established. S2: Map the exposure results of the verification points to a unified identifier, constrain the trusted corridor range volume to perform local alignment updates to form a new version of the trusted corridor range volume; classify the inconsistencies in the evidence layer into four types of conflicts: coordinate reference deviation, detection interpretation deviation, data missing, and path change, and label the impact domain; S3: Based on the corridor version chain, read the hierarchical corridor range volume set and generate a set of credible corridor hypotheses according to the source tags; for the action sequence, read the action envelope library, perform intersection and nearest distance determination between the action envelope and the corridor range volume corresponding to each credible corridor hypothesis, form a constraint trigger chain that records the action index, unified identifier, trigger type and separation distance of the first trigger, and output the feasibility label and the first trigger record under each hypothesis; construct stability index and verification value index and input them into the verification priority coefficient generator, output the verification priority coefficient for the verification candidate point, and then generate the minimum exploration closed loop point set and complete the action set splitting; S4: Read the stability gating result and verification priority coefficient. Based on the reference mapping table, write the actions with the reach indication quantity of reach and the associated unified identifier belonging to the verification trigger set into the reach uncertain corridor action set, and write the remaining actions into the non-reach alternative action set; freeze the reach uncertain corridor action set and use the completion of the verification event and the new version of the corridor as the release conditions, release the non-reach alternative action set, and archive the verification point, corridor version, gating criteria, action execution trajectory and deduction output into the twin version chain.

2. The simulation method for power transmission and transformation engineering construction based on digital twins according to claim 1, characterized in that, Step S1 includes: The construction survey control network was selected as the spatial reference benchmark. Rigid body alignment was performed based on the corresponding control points, and outliers were eliminated. Geometric normalization of the evidence layer was completed, and a mapping table between the unified identifier of underground objects and the evidence layer to the unified identifier was established.

3. The simulation method for power transmission and transformation engineering construction based on digital twins according to claim 2, characterized in that, Step S1 also includes: Based on the unified identifier, the center line is extracted from each evidence layer, a tubular credible corridor range volume is constructed, and the source label is registered. Based on the construction action sequence, action labels and action envelopes are generated. The reach indication and separation distance of the action envelope and the credible corridor range volume are calculated to form a reference mapping table, and the credible corridor range volume library and action envelope library are output.

4. The simulation method for power transmission and transformation engineering construction based on digital twins according to claim 3, characterized in that, Step S2 includes: Receive verification events and register verification points, exposure point sets, measurement method identifiers, and measurement resolution scales. Verification points are expressed using spatial reference benchmarks. Perform point inclusion retrieval and nearest distance retrieval based on the trusted corridor range library. Combine the mapping table from the evidence layer to the unified identifier to complete the unified identifier locking.

5. The simulation method for power transmission and transformation engineering construction based on digital twins according to claim 4, characterized in that, Step S2 also includes: The hierarchical centerline and radius function are updated according to the unified identifier and exposure point set by performing local alignment update rules, and a new version of the corridor is generated. Conflict classification is completed according to the consistency of residual direction and the difference in orientation topology, and the affected spatial domain is generated. The affected action reference domain is filtered according to the reference mapping table and written into the corridor version chain.

6. The simulation method for power transmission and transformation engineering construction based on digital twins according to claim 5, characterized in that, Step S3 includes: For each unified identifier, a baseline hypothesis is selected from the set of credible corridor hypotheses. The selection of the baseline hypothesis follows a parallel rule of prioritizing historical exposure evidence layers and the latest version index. When there is a conflict between the parallel rules, the source label confidence sequence is used to fix the selection order. Hypotheses other than the baseline hypothesis in the set of credible corridor hypotheses are used as the remaining hypotheses. The feasibility labels of the remaining hypotheses are compared with the feasibility labels of the baseline hypothesis one by one. The number of inconsistencies is counted, and the ratio of the number of inconsistencies to the number of comparisons is used to generate the feasibility flip ratio. A set of sensitive sections is generated on the construction layout line based on the first trigger record. A set of verification candidate points is generated around the sensitive sections by anchoring at the minimum net distance. The local alignment update rule is called to estimate the expected range volume reduction. The ratio of the expected range volume reduction to the volume represented by the verification workload is used as the corridor reduction benefit density.

7. The simulation method for power transmission and transformation engineering construction based on digital twins according to claim 6, characterized in that, Step S3 also includes: Based on the stability gating threshold, the feasible flip ratio is gating and screening is performed. A verification priority coefficient generator with monotonic constraints is used to map the feasible flip ratio and the narrow corridor benefit density to the verification priority coefficient for the verification candidate points, and the output is sorted. The minimum exploration closed loop point set is formed according to the coverage rule. According to the reference mapping table, the actions with the reach indication quantity and associated unified identifier belonging to the verification trigger set are written into the reach uncertain corridor action set, and the remaining actions are written into the non-reach alternative action set.

8. The method for simulating the construction of power transmission and transformation projects based on digital twins according to claim 7, characterized in that, Step S4 includes: Read the stability gating results and the minimum exploration closed-loop point set to generate a release unit record; bind the release unit record to the action identifier and the unified identifier set of the reach source, and associate it with the effective version index of the corridor version chain; for the reach of uncertain corridor action set, register the freeze judgment and release judgment; the release judgment corresponds to the verification event completion status and the new version of the corridor effective status.

9. The simulation method for power transmission and transformation engineering construction based on digital twins according to claim 8, characterized in that, Step S4 also includes: For actions that do not trigger alternative actions, allow them to proceed and archive the execution window and execution trajectory. For actions that trigger uncertain corridors, allow them to proceed and archive the execution window and execution trajectory when the release criteria are met. After the verification event is completed, trigger the new version of the corridor to take effect, filter the affected action set and perform a re-deduction and write-back. At the same time, write the freeze reason reference and the release basis reference into the version chain.

Images