Railway CRD tunneling process step distance intelligent optimization method and system

Abstract

The application discloses a kind of track CRD subsurface excavation process step distance intelligent optimization method and system, it is related to track subsurface excavation data processing technical field, including acquisition track CRD subsurface excavation process space-time data, generates process constraint sample and section state vector.Based on process constraint sample, CRD step distance constraint graph is constructed, and surrounding rock support coupling constraint quantity is calculated.According to surrounding rock support coupling constraint quantity, process step distance scheme is searched, and step distance optimization result and system execution instruction are generated.The application forms the closed-loop technical path of data sample-fig constraint-coupling quantity calculation-simulation check-instruction output under data processing framework, improves the executability, traceability and section state adaptability of track CRD subsurface excavation process step distance calculation.

Description

Technical Field

[0001] This invention relates to the field of track excavation data processing technology, specifically to a method and system for intelligent optimization of the step distance in track CRD excavation processes. Background Technology

[0002] With the development of digital acquisition of construction data for track CRD (Cut-Off Reduction) excavation and computer-aided decision-making technologies, the step distance of a work process is no longer merely a parameter for manual judgment based on construction experience. Instead, it is gradually being transformed into spatiotemporal sequence data, cross-sectional state data, process-dependent data, and constraint calculation data that can be processed by computing equipment. Existing technologies have begun to utilize database field mapping, graph data structures, state vector encoding, simulation result parsing, and constraint search algorithms to digitize the temporal sequence, spatial location, state changes, and execution conditions between different work process nodes during CRD excavation. Furthermore, they generate work step distance analysis results within a data processing, model calculation, graph structure traversal, and optimization solution framework, providing a computer-based foundation for the intelligent optimization of work step distances in track CRD excavation.

[0003] However, existing methods for processing track CRD (Cut-off Reduction) process step distance data mainly remain at the stage of monitoring data recording, empirical rule input, or single calculation result output. It is difficult to uniformly encode the data fields corresponding to the upper left guide tunnel, upper right guide tunnel, lower left guide tunnel, lower right guide tunnel, temporary central partition wall, initial support, and secondary lining into computational objects with node attributes, edge attributes, and status labels. Although existing multi-source data fusion methods can complete the collection of sensor data, process logs, and cross-section model data, they usually lack sample construction rules for CRD process step distance optimization, resulting in the input data still requiring manual interpretation and format conversion before entering the calculation model. Existing constraint optimization algorithms mostly treat process step distance as ordinary numerical variables and do not synchronously express the process preceding relationships, cross-section spatial relationships, support closure status, monitoring response changes, and verification acquisition trigger conditions in the computer graph structure, making it difficult to form a traceable constraint calculation path in the search process. Existing simulation or prediction results are also usually saved as independent output results and are not converted into constraint quantities, state vectors, or system execution instructions that can be directly called in the next round of step distance search. Therefore, existing technologies struggle to achieve closed-loop calculations within a data processing framework, from process spatiotemporal data acquisition, constraint sample construction, coupled constraint quantity calculation to step size optimization result generation. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by this invention is that the existing intelligent optimization method for step distance in track CRD underground excavation process has the following problems: the process spatiotemporal data is difficult to encode into computer-processable constraint samples, the cross-sectional state data is difficult to convert into traversable graph structure constraint quantities, the step distance search results are difficult to update in a closed loop based on computational feedback, and there is a problem of how to generate traceable, verifiable and output-to-execution system step distance optimization results under the data processing framework.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a method for intelligent optimization of step distance in track CRD underground excavation process, including collecting spatiotemporal data of track CRD underground excavation process and generating process constraint samples and cross-sectional state vectors.

[0007] Based on the process constraint samples, a CRD step distance constraint diagram is constructed, and the coupling constraint amount of the surrounding rock support is calculated.

[0008] Based on the coupling constraints of the surrounding rock support, the step distance scheme of the process is searched, and the step distance optimization results and system execution instructions are generated.

[0009] As a preferred embodiment of the intelligent optimization method for the step distance of the track CRD tunneling process described in this invention, the acquisition of spatiotemporal data for the track CRD tunneling process includes: establishing a process coding table according to the construction sequence of the upper left guide tunnel, upper right guide tunnel, lower left guide tunnel, lower right guide tunnel, temporary central partition wall, initial support, and secondary lining of the track CRD tunneling section; writing the face mileage, excavation outline, advanced support type, temporary support closure time, initial support closure time, monitoring section number, surrounding rock grade, arch settlement, perimeter convergence, surface settlement, steel arch strain, and shotcrete age corresponding to each construction cycle into a unified data frame; performing timestamp alignment, mileage coordinate normalization, missing item marking, and anomaly item isolation on the unified data frame, retaining the source identifier of the anomaly item, and not deleting the construction cycle corresponding to the anomaly item; mapping sensor monitoring data, three-dimensional geological model data, and structured construction log data to the same CRD section coordinate system to obtain spatiotemporal data of the track CRD tunneling process containing spatial location, process status, and support status.

[0010] As a preferred embodiment of the intelligent optimization method for the step distance of the track CRD tunnel excavation process described in this invention, the generation of process constraint samples and cross-section state vectors includes: using the state change of a single track CRD tunnel cross-section between adjacent construction cycles as a sample slice; and using the following parameters from the sample slices as process step distance variables: the excavation lag of the upper left pilot tunnel, the excavation lag of the upper right pilot tunnel, the excavation lag of the lower left pilot tunnel, the excavation lag of the lower right pilot tunnel, the closure lag of the temporary central partition wall, the closure lag of the initial support, and the follow-up lag of the secondary lining. The following parameters from the sample slices are used as cross-section state variables: surrounding rock grade code, monitored deformation increment, deformation rate, support strain increment, support age, number of completed pilot tunnels, and length of unclosed support. Based on construction sequence constraints, space occupancy constraints, support age constraints, and monitored deformation constraints, constraint labels are generated for each sample slice. These constraint labels include allowed advancement, restricted advancement, support priority, and verification collection. The process constraint samples and cross-section state vectors are obtained by concatenating the fixed fields of the process step distance variables, cross-section state variables, and constraint labels.

[0011] As a preferred embodiment of the intelligent optimization method for the step distance of the track CRD tunnel excavation process described in this invention, the step distance constraint graph constructed based on process constraint samples includes setting four pilot tunnel excavation nodes, temporary intermediate partition wall nodes, initial support nodes, invert arch closure nodes, and secondary lining nodes in the track CRD tunnel excavation section as graph nodes, and setting the temporal dependency of adjacent processes, spatial compatibility of the same section, deformation transmission of adjacent sections, closure dependency of support structures, and feedback dependency of monitoring points on the construction cycle as directed edges. The minimum step distance, maximum step distance, closure waiting time, spatial occupancy status, deformation transmission direction, and monitoring trigger status extracted from the process constraint samples are written into each directed edge. When the unclosed support length in the section state vector increases or the monitored deformation increment continues to increase, the attributes of the directed edges connected to the temporary intermediate partition wall node, initial support node, and adjacent pilot tunnel excavation nodes are updated. When the construction cycle completes support closure and the monitoring data enters a stable update interval, the constraint marker of the corresponding directed edge is removed. By traversing the graph, the set of preceding processes, the set of mutually exclusive processes, and the set of monitoring constraints for each candidate advancement process are obtained, thus generating the CRD step distance constraint graph.

[0012] As a preferred embodiment of the intelligent optimization method for the step distance of the track CRD tunneling operation described in this invention, the calculation of the surrounding rock support coupling constraint includes: reading the monitoring constraint set and support closure dependency edge corresponding to the current candidate advancement operation from the CRD step distance constraint diagram; calculating the deviation of the monitored deformation increment relative to the historical stable interval, the change of the deformation rate relative to the current cross-section support age, the deviation of the support strain increment relative to the same surrounding rock grade sample, and the amplification of the unclosed support length relative to the candidate advancement step distance. The deviation, change, support strain deviation, and amplification are adaptively normalized according to the surrounding rock grade, support age, and pilot tunnel location in the cross-section state vector to generate local constraint components corresponding to the candidate advancement operation. When the local constraint component and the time-series dependency edge in the CRD step distance constraint diagram both point to the same pilot tunnel excavation node, the candidate advancement operation is marked as requiring a shortened step distance. When the local constraint component is only associated with the support closure dependency edge, the candidate advancement operation is marked as having support closure priority. All local constraint components are aggregated to the corresponding cross-section according to the directed edge propagation direction to obtain the surrounding rock support coupling constraint.

[0013] As a preferred embodiment of the intelligent optimization method for the step distance of the track CRD tunnel excavation process described in this invention, the step distance search based on the surrounding rock support coupling constraint includes: reading the minimum step distance, maximum step distance, set of preceding processes, and set of mutually exclusive processes for each candidate advancement process from the CRD step distance constraint diagram to form a candidate set of offset step distances. Feasibility screening is performed on the candidate set of offset step distances based on the surrounding rock support coupling constraint, eliminating candidate step distances that violate the set of preceding processes, temporarily storing candidate step distances that trigger the set of mutually exclusive processes as conflict candidates, and retaining candidate step distances that satisfy temporal dependency, spatial compatibility, and monitoring constraints as feasible candidates. Conflict relaxation calculations are performed on the conflict candidates, adjusting the value range of the candidate step distances in the order of priority for support closure, deformation stability, and symmetrical advancement of the pilot tunnel. Feasible candidates are input into a digital section simulation verification model, which outputs the deformation prediction, support closure status, and space occupancy status for the next construction cycle. The candidate step distances that pass the simulation verification are iteratively sorted to obtain the process step distance scheme.

[0014] As a preferred embodiment of the intelligent optimization method for the step distance of the track CRD tunnel excavation process described in this invention, the generation of step distance optimization results and system execution instructions includes converting the process step distance scheme into a process step distance table for the track CRD tunnel excavation section. The process step distance table includes the pilot tunnel number, start and end mileage, target step distance, support closure sequence, allowed advancement marker, verification acquisition marker, and next acquisition trigger condition. When the target step distance of any pilot tunnel in the process step distance scheme is shortened by conflict relaxation calculation, the directed edge number and section state vector field corresponding to the shortening reason are written into the process step distance table. When the deformation prediction output by the digital section simulation verification model exceeds the stable update interval of the same surrounding rock grade, a verification acquisition instruction is generated, but no advancement instruction is generated. When the deformation prediction, support closure state, and space occupancy state all satisfy the edge attributes corresponding to the CRD step distance constraint diagram, the process step distance table is encapsulated into a system execution instruction. The system execution instruction is sent to the data acquisition module, simulation verification module, and field terminal display module, and the instruction version number, generation time, and corresponding section state vector are recorded.

[0015] As a preferred embodiment of the intelligent optimization system for step distance in the track CRD underground excavation process described in this invention, it includes a spatiotemporal sample construction module, a constraint map quantity calculation module, and a step distance optimization execution module.

[0016] The spatiotemporal sample construction module is used to collect spatiotemporal data of the track CRD underground excavation process and generate process constraint samples and cross-sectional state vectors.

[0017] The constraint diagram quantity calculation module is used to construct a CRD step distance constraint diagram based on the process constraint sample and calculate the coupling constraint quantity of the surrounding rock support.

[0018] The step distance optimization execution module is used to search for process step distance schemes based on the coupling constraints of the surrounding rock support, and generate step distance optimization results and system execution instructions.

[0019] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement a method for intelligent optimization of the step distance in the track CRD underground excavation process.

[0020] A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of a method for intelligent optimization of the step distance in a track CRD (Cross-Drilling) process.

[0021] The beneficial effects of this invention are: Within the data processing framework, the excavation, support, monitoring, and geological data of the track CRD cut-and-cover tunnel are first converted into a unified indexable data structure. Then, process constraint samples and cross-sectional state vectors are formed based on the differences in construction cycles. This eliminates calculation breakpoints caused by different sources, time scales, and spatial coordinates from the data layer, providing clear input for subsequent constraint map construction and step-by-step optimization search.

[0022] The problem of step distance in the track CRD (Cut-Off Relief) excavation process is transformed into a computable directed heterogeneous graph constraint problem. Then, monitored deformation, support strain, and unclosed support length are integrated into a surrounding rock support coupling constraint. This approach does not involve a simple comparison of single monitoring thresholds; instead, it uses graph edge attributes and coupling constraints to jointly limit candidate step distances, allowing adjacent processes, adjacent sections, and support closure states to participate in the calculation within the same algorithm structure.

[0023] In the executable algorithm, graph constraint filtering, conflict relaxation, digital cross-section simulation verification, and projection iterative sorting are connected into a closed loop, so that the output is not only a single step distance value, but also a structured instruction containing the target step distance, triggering reason, verification conditions, and system execution interface. This processing can solve the problem that existing single threshold judgment methods cannot simultaneously handle process pre-processing, spatial compatibility, surrounding rock deformation, and support closure feedback. Attached Figure Description

[0024] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is an overall flowchart of a method for intelligent optimization of step distance in track CRD underground excavation, as provided in Embodiment 1 of the present invention. Detailed Implementation

[0026] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0027] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for intelligent optimization of step distance in track CRD underground excavation is provided, comprising: S1: Collect spatiotemporal data of the track CRD underground excavation process and generate process constraint samples and cross-sectional state vectors.

[0028] A process coding table was established according to the construction sequence of the left upper pilot tunnel, right upper pilot tunnel, left lower pilot tunnel, right lower pilot tunnel, temporary central partition wall, initial support, and secondary lining of the track CRD cut-and-cover section. The working face mileage, excavation outline, advanced support type, temporary support closure time, initial support closure time, monitoring section number, surrounding rock grade, arch settlement, perimeter convergence, surface settlement, steel arch strain, and shotcrete age corresponding to each construction cycle were written into a unified data frame. Timestamp alignment, mileage coordinate normalization, missing item marking, and anomaly isolation were performed on the unified data frame, retaining the source identifier of anomalies and not deleting the construction cycle corresponding to the anomalies. Sensor monitoring data, 3D geological model data, and structured construction log data were mapped to the same CRD section coordinate system to obtain spatiotemporal data of the track CRD cut-and-cover process, including spatial location, process status, and support status.

[0029] Furthermore, the process coding table adopts a five-field primary key: section number, pilot tunnel number, process number, construction cycle number, and data source number. Each field is stored as an integer code. The section number is generated by increasing the track mileage from small mileage to large mileage. The pilot tunnel number is generated in a fixed order: upper left pilot tunnel, upper right pilot tunnel, lower left pilot tunnel, and lower right pilot tunnel. The process number is generated in a fixed order: excavation, installation of temporary central partition wall, initial support, invert arch closure, and secondary lining. This allows the spatial and temporal data of the track CRD tunnel excavation process to be directly indexed by the data processing program.

[0030] Furthermore, timestamp alignment is based on the start time of the construction cycle, converting the sampling time of monitoring equipment, the completion time of support construction, and the entry time of construction logs into relative minutes. Mileage coordinate normalization uses the current cross-section design center mileage as zero, converting the working face mileage, monitoring point mileage, and support closure mileage into the same axial coordinate system. When multiple monitoring sampling values ​​exist within the same construction cycle, the sampling value with the sampling time closest to the support closure time is retained, and the remaining sampling values ​​are written to the history sequence field.

[0031] Furthermore, missing item marking is not performed by deleting samples. The value of the missing item marking field is determined by the data source. Missing items caused by sensor communication interruption are marked as acquisition missing, missing items caused by failure to enter construction logs are marked as record missing, and missing items caused by failure to cover the 3D geological model are marked as model missing. Anomaly isolation is identified by the median difference of three consecutive sampling cycles at the same monitoring point. After anomalies enter the isolation field, they remain bound to the original construction cycle for reference in subsequent review and acquisition instructions.

[0032] Furthermore, when the three-dimensional geological model data is mapped to the CRD cross-section coordinate system, the stratigraphic boundaries, groundwater distribution, surrounding rock grade, and distances to existing structures intersecting with the cross-section centerline are extracted along the line axis. The spatial positions of the arch crown, arch waist, sidewalls, and invert are extracted vertically. The above spatial positions are associated with the monitoring cross-section number to form a structured array that can be input into the calculation model.

[0033] Using the state changes of a single track CRD cut-and-cover section between adjacent construction cycles as sample slices, the following parameters are used as process step distance variables: the excavation lag of the upper left pilot tunnel, the upper right pilot tunnel, the lower left pilot tunnel, the lower right pilot tunnel, the closure lag of the temporary central partition wall, the closure lag of the initial support, and the follow-up lag of the secondary lining. The surrounding rock grade code, monitored deformation increment, deformation rate, support strain increment, support age, number of completed pilot tunnels, and length of unclosed support in the sample slices are used as section state variables. Constraint labels are generated for each sample slice based on construction sequence constraints, space occupancy constraints, support age constraints, and monitored deformation constraints. These labels include allowed advancement, restricted advancement, support priority, and verification data collection. The process constraint samples and section state vectors are obtained by concatenating the fixed fields of the process step distance variables, section state variables, and constraint labels.

[0034] Furthermore, the starting boundary of the sample slice is the time when the support status of the previous construction cycle is written, and the ending boundary is the time when the monitoring data of the current construction cycle is aligned. The excavation lag within the sample slice is calculated by the difference between the target tunnel face mileage and the adjacent tunnel face mileage. The support closure lag is calculated by the difference between the target support closure mileage and the corresponding tunnel face mileage. The secondary lining follow-up lag is calculated by the difference between the secondary lining end mileage and the initial support closure mileage.

[0035] Furthermore, the monitored deformation increment is the difference between the monitoring value of the current construction cycle and the monitoring value of the previous construction cycle; the deformation rate is the monitored deformation increment divided by the relative number of minutes between two construction cycles; the support strain increment is the difference between the current strain value of the steel arch frame and the strain reference value at the time of support closure; and the support age is the relative number of hours from the completion time of shotcrete to the end boundary of the current sample slice. All of the above variables are bound to the section number, the pilot tunnel number, and the process number.

[0036] Furthermore, the generation of constraint labels does not rely on manually written rule text, but is generated by a fixed logic program: When all preceding procedures are completed, space occupancy status is not conflicting, and the monitored deformation increment enters a stable update interval, "allow progress" is written. When preceding procedures are not completed or the unclosed support length exceeds the corresponding pilot tunnel candidate step distance, "restrict progress" is written. When the support age has not reached the lower limit of the stable age of historical samples of the same surrounding rock grade, "support priority" is written. When the missing item marker or the anomaly isolation field is set, "review collection" is written.

[0037] Furthermore, the cross-section state vector is generated according to a fixed field order: process step distance variable first, cross-section state variable in the middle, and constraint label last. The field length, version number, and sampling source verification code are written at the beginning of the vector. When subsequent data acquisition fields are added, extended fields are only appended to the end of the vector without changing the positions of existing fields, ensuring that the CRD step distance constraint diagram and the digital cross-section simulation verification model can read data input in the same format.

[0038] It should be noted that the design idea of ​​this step is to first convert the excavation, support, monitoring and geological data of the track CRD tunnel into a unified indexable data structure under the data processing framework, and then form process constraint samples and cross-section state vectors based on the differences in construction cycles. This eliminates the calculation breakpoints caused by different sources, different time scales and different spatial coordinates from the data layer, and provides clear input for subsequent constraint map construction and step-by-step optimization search.

[0039] S2: Construct a CRD step distance constraint diagram based on process constraint samples and calculate the coupling constraint amount of surrounding rock support.

[0040] Four nodes in the track CRD cut-and-cover section—the pilot tunnel excavation node, the temporary central partition wall node, the initial support node, the invert arch closure node, and the secondary lining node—are set as graph nodes. The temporal dependencies between adjacent processes, the spatial compatibility of the same section, the deformation transfer between adjacent sections, the closure dependency of the support structure, and the feedback dependency of monitoring points on the construction cycle are set as directed edges. The minimum step distance, maximum step distance, closure waiting time, spatial occupancy status, deformation transfer direction, and monitoring trigger status, extracted from the process constraint samples, are written into each directed edge. When the unclosed support length in the section state vector increases or the monitored deformation increment continues to increase, the attributes of the directed edges connected to the temporary central partition wall node, the initial support node, and the adjacent pilot tunnel excavation node are updated. When the construction cycle completes support closure and the monitoring data enters a stable update interval, the constraint marker of the corresponding directed edge is removed. The set of preceding processes, the set of mutually exclusive processes, and the set of monitoring constraints for each candidate advancement process are obtained through graph traversal, resulting in the CRD step distance constraint graph.

[0041] Furthermore, the CRD step-constraint graph is stored using a directed heterogeneous graph structure. The type fields of the graph nodes include pilot tunnel excavation, temporary intermediate diaphragm wall, initial support, invert arch closure, and secondary lining. The type fields of the directed edges include temporally dependent edges, spatially compatible edges, deformation transfer edges, closure dependent edges, and monitoring feedback edges. Each graph node is bound to a section number, pilot tunnel number, and process number. Each directed edge is bound to a start node, end node, edge type, edge attribute array, and constraint flags, enabling the graph structure to be traversed and updated by a computer program.

[0042] Furthermore, the minimum and maximum step distances are not fixed constants, but are generated from historical allowable advancement samples of the same surrounding rock grade, the same pilot tunnel location, and the same support condition within the process constraint sample. The closure waiting time is jointly determined by the shotcrete age, the strain increment of the steel arch frame, and the closure time of the temporary central diaphragm wall. The space occupancy status includes the relative position of the pilot tunnel face, the location of temporary supports, and the accessible boundaries for equipment. The space occupancy status is only used to determine geometric compatibility and is not used for construction resource management.

[0043] Furthermore, the stable update interval is generated by the crown settlement increment, perimeter convergence increment, and surface settlement increment of the same monitoring point in the most recent six effective construction cycles. The upper limit of the stable update interval is the sum of the absolute value of the median of the above six effective construction cycles and the interquartile range. When there are fewer than six effective construction cycles, the stable update interval is temporarily supplemented by stable samples from adjacent sections of the same surrounding rock grade, and the supplementary source is written into the directed edge attribute array.

[0044] Furthermore, the graph traversal uses a reverse tracing method from the target candidate advancement node to determine the set of preceding processes, uses the constraint marking of spatially compatible edges within the same cross-section to determine the set of mutually exclusive processes, and uses the monitoring feedback edges connected to the target candidate advancement node to determine the set of monitoring constraints. The set of preceding processes, the set of mutually exclusive processes, and the set of monitoring constraints are all output as an array of node numbers for use in the offset distance candidate set selection of S3.

[0045] The monitoring constraint set and support closure dependency edge corresponding to the current candidate advancement procedure are read from the CRD step distance constraint diagram. The deviation of the monitored deformation increment from the historical stable interval, the change in deformation rate relative to the current support age, the deviation of the support strain increment relative to samples of the same surrounding rock grade, and the amplification of the unclosed support length relative to the candidate advancement step distance are calculated. The deviation, change, support strain deviation, and amplification are adaptively normalized according to the surrounding rock grade, support age, and pilot tunnel location in the section state vector to generate local constraint components corresponding to the candidate advancement procedure. When the local constraint component and the time-series dependency edge in the CRD step distance constraint diagram both point to the same pilot tunnel excavation node, the candidate advancement procedure is marked as requiring a shortened step distance. When the local constraint component is only associated with the support closure dependency edge, the candidate advancement procedure is marked as having support closure priority. All local constraint components are aggregated to the corresponding section according to the directed edge propagation direction to obtain the surrounding rock support coupling constraint quantity.

[0046] Furthermore, the deviation of the monitored deformation increment is generated by the difference between the monitored deformation increment of the current construction cycle and the upper limit of the historical stable interval; the change in deformation rate is generated by the difference between the current deformation rate and the historical rate envelope corresponding to the same cross-section support age; the deviation of support strain is generated by the difference between the strain increment of the steel arch frame and the strain stability envelope of the same surrounding rock grade sample; and the amplification of the unclosed support length is generated by the ratio between the unclosed support length and the candidate advance step distance.

[0047] Furthermore, to ensure that the coupling constraint of the surrounding rock support can be repeatedly executed in the calculation program, this embodiment first uses the following formula to calculate the section numbering. The normalized component of the monitored deformation, the fraction in the formula represents the dimensionless position of the current monitored deformation state relative to the stability datum and the limiting boundary of the same cross section.

[0048]

[0049] in, The section number is indicated as The normalized component of the monitored deformation. The section number is indicated as The monitored deformation increment. The section number is indicated as The stable update interval center value. This represents the upper limit of the stable update interval for section number h.

[0050] The normalized component of deformation rate, the normalized component of support strain, and the normalized component of unclosed support length are generated using the same stability datum and confined boundary structure. These three types of normalized components are denoted as follows: , and .

[0051] Furthermore, the coupling constraint of the surrounding rock support is aggregated and monitored using the following formulas: deformation normalization component, deformation rate normalization component, support strain normalization component, and unclosed support length normalization component.

[0052]

[0053] in, The section number is indicated as The coupling constraint of the surrounding rock support. , , and These represent the normalization coefficients corresponding to the monitored deformation increment, deformation rate, support strain increment, and unclosed support length, respectively. The four normalization coefficients are generated efficiently from the fields of historical samples of the same surrounding rock grade and remain unchanged within the same construction cycle. The section number is indicated as The normalized component of the deformation rate, The section number is indicated as The normalized component of the support strain, The section number is indicated as The normalized component of the unclosed support length.

[0054] Furthermore, when the calculated result of the surrounding rock support coupling constraint falls within the allowable advancement range, the CRD step distance constraint diagram retains the maximum step distance edge attribute of the candidate advancement process. When the calculation result falls within the restricted advancement range, the CRD step distance constraint diagram updates the maximum step distance edge attribute of the candidate advancement process to the smaller value between the current candidate advancement step distance and the unclosed support length. When the calculation result falls within the verification and acquisition range, the CRD step distance constraint diagram sets the restriction mark of the monitoring feedback edge.

[0055] Furthermore, the boundaries of the permitted advance interval, restricted advance interval, and verification collection interval are adaptively generated from historical stable samples. The upper boundary of the permitted advance interval is taken as the upper quartile value of the surrounding rock support coupling constraint of historical permitted advance samples of the same surrounding rock grade. The lower boundary of the verification collection interval is taken as the lower quartile value of the surrounding rock support coupling constraint of historical verification collection samples of the same surrounding rock grade. The restricted advance interval lies between the above two boundaries. Boundary values ​​are written into the versioned parameter table after secondary lining is completed at each cross-section.

[0056] Furthermore, the feasible domain of the step distance for candidate advancement processes is jointly defined by the lower bound of the candidate step distance, the upper bound of the candidate step distance, the allowable boundary of the surrounding rock support coupling constraint, and the spatial compatibility judgment result. When the output of the spatial compatibility judgment function is that the constraint flag has not been triggered, the candidate step distance remains in the offset step distance candidate set. When the output of the spatial compatibility judgment function is that the constraint flag has been triggered, the candidate step distance enters the conflict candidate set and waits for the conflict relaxation calculation of S3.

[0057] It should be noted that the design concept of this step is to transform the step distance problem of the track CRD cut-and-cover excavation process into a computable directed heterogeneous graph constraint problem, and then integrate the monitored deformation, support strain, and unclosed support length into the surrounding rock support coupling constraint quantity. This processing method is not a simple comparison of a single monitoring threshold, but rather uses graph edge attributes and coupling constraint quantities to jointly limit the candidate step distance, so that adjacent processes, adjacent sections, and support closure states can participate in the calculation within the same algorithm structure.

[0058] S3: Based on the coupling constraint of the surrounding rock support, search for the process step distance scheme, generate the step distance optimization results and system execution instructions.

[0059] The minimum step distance, maximum step distance, set of preceding procedures, and set of mutually exclusive procedures are read from the CRD step distance constraint diagram to form a candidate set of offset step distances. Feasibility screening is performed on the candidate set of offset step distances based on the surrounding rock support coupling constraints. Candidate step distances that violate the set of preceding procedures are eliminated, candidate step distances that trigger the set of mutually exclusive procedures are temporarily stored as conflict candidates, and candidate step distances that satisfy temporal dependency, spatial compatibility, and monitoring constraints are retained as feasible candidates. Conflict relaxation calculations are performed on the conflict candidates, adjusting the value range of the candidate step distances in the order of support closure priority, deformation stability priority, and symmetrical advancement of the pilot tunnel priority. Feasible candidates are input into the digital section simulation verification model, which outputs the deformation prediction, support closure status, and space occupancy status for the next construction cycle. The candidate step distances that pass the simulation verification are iteratively sorted to obtain the procedure step distance scheme.

[0060] Furthermore, the step size of the candidate distance distance set is generated by the process step distance variable field of the current pilot tunnel. The lower bound of the candidate step distance is the minimum step distance in the CRD step distance constraint diagram, and the upper bound of the candidate step distance is the maximum step distance in the CRD step distance constraint diagram. When the maximum step distance is updated due to the surrounding rock support coupling constraint of S2, the candidate distance distance set is immediately regenerated according to the updated maximum step distance, avoiding the use of the candidate boundary of the previous construction cycle.

[0061] Furthermore, feasibility screening is performed in the following order: the set of prerequisite processes, the set of mutually exclusive processes, and the set of monitoring constraints. The set of prerequisite processes is used to determine whether the target candidate process has the entry conditions. The set of mutually exclusive processes is used to determine whether the target candidate process has a spatial conflict with other pilot tunnel excavations or support closures on the same cross section. The set of monitoring constraints is used to determine whether the target candidate process has verification and data collection markers. All three types of judgments use the node numbers and directed edge attributes in the CRD step distance constraint diagram as input.

[0062] Furthermore, the conflict relaxation calculation does not directly delete conflict candidates, but rather reallocates the range of candidate step distances without disrupting the set of preceding processes. When the conflict source originates from a support closure dependency edge, the candidate step distance for excavation nodes is shortened first, while the support nodes are retained. When the conflict source originates from a monitoring feedback edge, candidate tunnel advancements with stable deformation are retained first. When the conflict source originates from a spatially compatible edge, the candidate step distances are adjusted in pairs according to the symmetrical relationship between the upper left and upper right tunnels, and between the lower left and lower right tunnels.

[0063] Furthermore, the digital cross-section simulation verification model includes a data input layer, a constraint encoding layer, a cross-section response prediction layer, and a verification output layer. The data input layer receives the cross-section state vector, the adjacency list of the CRD step constraint graph, and the candidate set of offset step distances. The constraint encoding layer encodes node types, edge types, and constraint tags into a sparse matrix. The cross-section response prediction layer employs a concatenated structure of graph convolutional units and gated loop units. The graph convolutional units read the edge attributes of adjacent processes and adjacent cross-sections, while the gated loop units read the monitoring sequence of continuous construction cycles. The verification output layer outputs the deformation prediction, support closure status, and space occupancy status for the next construction cycle.

[0064] Furthermore, when ranking the candidate step sizes verified through simulation, the candidate scheme evaluation metric is used to number the candidate schemes. The candidate step size combinations are calculated, and the smaller the evaluation quantity of the candidate scheme, the higher the ranking of the candidate step size combination.

[0065]

[0066] in, Indicates that the candidate scheme number is Evaluation of candidate solutions. , , and These represent the evaluation coefficients corresponding to coupling constraints, simulation deviations, verification marks, and step boundary occupancy, respectively. The evaluation coefficients are generated by inversely from the field integrity rate and simulation error of the most recent effective construction cycle for the same cross-section. Indicates the candidate scheme number is The maximum surrounding rock support coupling constraint amount, Indicates the candidate scheme number is The simulated deformation deviation. Indicates that the candidate scheme number is The number of verification and collection markers, Indicates that the candidate scheme number is The amount of space occupied by the step margin boundary.

[0067] Furthermore, the iterative sorting process writes the top-ranked candidate step distance combination into the current iteration vector and rewrites the deformation prediction output by the digital cross-section simulation verification model into the cross-section state vector. If the recalculated surrounding rock support coupling constraint is still within the allowable advancement range, the current candidate step distance combination is output. If the recalculated surrounding rock support coupling constraint enters the restricted advancement range, a new shortening direction candidate set is generated centered on the current candidate step distance combination.

[0068] Furthermore, in this embodiment, the following projection update formula is used to ensure that the candidate step size combination is always within the step size feasible region formed by S2.

[0069]

[0070] in, Indicates the first The candidate step vector obtained from the iteration. No. The candidate step vector used in the next iteration. Represents the feasible region by step size Projection calculation operator, step size feasible region It is composed of the lower limit of candidate step distance, the upper limit of candidate step distance, the allowable boundary of surrounding rock support coupling constraint, and the spatial compatibility judgment results of all candidate propulsion procedures. Indicates the first The update step size for each iteration is automatically scaled by the difference in evaluation values ​​between two adjacent candidate solutions. Indicates the first The discrete difference direction of the candidate scheme evaluation quantity with respect to the candidate step vector in the next iteration.

[0071] The process step distance scheme is converted into a process step distance table for the track CRD cut-and-cover section. This table includes the pilot tunnel number, start and end mileage, target step distance, support closure sequence, allowable advance marker, verification acquisition marker, and next acquisition trigger condition. When the target step distance of any pilot tunnel in the process step distance scheme is shortened by conflict relaxation calculation, the directed edge number and section state vector field corresponding to the shortening reason are written into the process step distance table. When the deformation prediction output by the digital section simulation verification model exceeds the stable update interval of the same surrounding rock grade, a verification acquisition instruction is generated, but no advance instruction is generated. When the deformation prediction, support closure state, and space occupancy state all satisfy the edge attributes corresponding to the CRD step distance constraint diagram, the process step distance table is encapsulated into a system execution instruction. This system execution instruction is sent to the data acquisition module, simulation verification module, and field terminal display module, and the instruction version number, generation time, and corresponding section state vector are recorded.

[0072] Furthermore, the process step table is stored in a machine-readable field format, in which the pilot tunnel number and start and end mileage are used to locate the target advancement section, the target step distance is used to write the candidate advancement length for the next construction cycle, the support closure sequence is used to write the sequential relationship between the temporary central partition wall, the initial support and the invert arch closure, the advance permission mark is used to control the advancement prompts of the on-site terminal display module, and the verification acquisition mark is used to control the data acquisition module to add monitoring samples.

[0073] Furthermore, the directed edge numbers corresponding to the shortening causes include temporally dependent edge numbers, spatially compatible edge numbers, deformation transfer edge numbers, closure dependent edge numbers, and monitoring feedback edge numbers. The cross-sectional state vector fields include the monitoring deformation increment, deformation rate, support strain increment, support age, and unclosed support length that triggered the shortening. The system executes instructions to write the above fields into the instruction traceability area for the process constraint sample generation program of the next construction cycle to read.

[0074] Furthermore, the verification acquisition command includes the monitoring point number, acquisition time interval, sensor type, verification section number, and verification field name. The advance command includes the target pilot tunnel number, target step distance, allowable advance time window, support closure preconditions, and command failure conditions. The two types of commands are mutually exclusive. When the verification acquisition command is generated, the advance command field remains empty to prevent the system from executing commands that simultaneously express the two opposite states of verification and advance.

[0075] Furthermore, the system execution instructions are jointly read by the data acquisition module, constraint diagram generation module, simulation verification module, step distance optimization module, and field terminal display module. The data acquisition module updates the spatiotemporal data of the track CRD tunneling process according to the verification acquisition instructions; the constraint diagram generation module updates the CRD step distance constraint diagram according to the instruction traceability area; the simulation verification module triggers the next round of cross-sectional response prediction based on the target step distance; the step distance optimization module determines whether to overwrite the old results based on the instruction version number; and the field terminal display module only displays the technical execution information corresponding to the allowable advancement marker and the verification acquisition marker.

[0076] Furthermore, the version number of the system execution command is generated by concatenating the section number, construction cycle number, and iteration number. When the same section repeatedly generates step distance optimization results within the same construction cycle, only the system execution command with the largest iteration number is retained, and the old version command is written to the historical command table. The historical command table does not participate in the next round of step distance search; it is only used to maintain the traceability of the data processing process.

[0077] It should be noted that the design concept of this step is to connect graph constraint filtering, conflict relaxation, digital cross-section simulation verification, and projection iterative sorting into a closed loop within the executable algorithm. This ensures that the output is not just a single step distance value, but a structured instruction containing the target step distance, triggering cause, verification conditions, and system execution interface. This processing can solve the problem that existing single-threshold judgment methods cannot simultaneously handle pre-process considerations, spatial compatibility, surrounding rock deformation, and support closure feedback.

[0078] Example 2, an embodiment of the present invention, provides an intelligent optimization system for the step distance in the track CRD underground excavation process, including a spatiotemporal sample construction module, a constraint map quantity calculation module, and a step distance optimization execution module.

[0079] The spatiotemporal sample construction module is used to collect spatiotemporal data of the track CRD underground excavation process and generate process constraint samples and cross-sectional state vectors.

[0080] The constraint diagram quantity calculation module is used to construct a CRD step distance constraint diagram based on process constraint samples and calculate the coupling constraint quantity of surrounding rock support.

[0081] The step distance optimization execution module is used to search for step distance schemes based on the coupling constraints of the surrounding rock support, and generate step distance optimization results and system execution instructions.

Claims

1. A method for intelligent optimization of step distance in track CRD underground excavation, characterized in that, include: Collect spatiotemporal data of the track CRD underground excavation process to generate process constraint samples and cross-sectional state vectors; Based on the process constraint samples, a CRD step distance constraint diagram is constructed, and the coupling constraint amount of surrounding rock support is calculated. Based on the coupling constraints of the surrounding rock support, the step distance scheme of the process is searched, and the step distance optimization results and system execution instructions are generated.

2. The intelligent optimization method for the step distance in the track CRD underground excavation process as described in claim 1, characterized in that: The spatiotemporal data collected during the track CRD tunneling process include... According to the construction sequence of the left upper guide tunnel, right upper guide tunnel, left lower guide tunnel, right lower guide tunnel, temporary central partition wall, initial support, and secondary lining of the track CRD cut-and-cover section, a process coding table is established. The working face mileage, excavation outline, advanced support type, temporary support closure time, initial support closure time, monitoring section number, surrounding rock grade, arch settlement, perimeter convergence, surface settlement, steel arch strain, and shotcrete age corresponding to each construction cycle are written into a unified data frame. For unified data frames, perform timestamp alignment, mileage coordinate normalization, missing item marking and abnormal item isolation, retain the source identifier of abnormal items, and do not delete the construction cycle corresponding to abnormal items; By mapping sensor monitoring data, 3D geological model data, and structured construction log data to the same CRD cross-section coordinate system, spatiotemporal data of the track CRD tunneling process, including spatial location, process status, and support status, are obtained.

3. The intelligent optimization method for the step distance in the track CRD underground excavation process as described in claim 2, characterized in that: The generation process constraint samples and cross-sectional state vectors include, The state changes of a single track CRD cut-and-cover section between adjacent construction cycles are used as sample slices. The excavation lag of the upper left guide tunnel, the excavation lag of the upper right guide tunnel, the excavation lag of the lower left guide tunnel, the excavation lag of the lower right guide tunnel, the closure lag of the temporary middle partition wall, the closure lag of the initial support, and the follow-up lag of the secondary lining in the sample slices are used as process step variables. The surrounding rock grade code, monitored deformation increment, deformation rate, support strain increment, support age, number of completed pilot tunnels and length of unclosed support in the sample slice are used as cross-sectional state variables. Based on construction sequence constraints, space occupation constraints, support age constraints, and monitoring deformation constraints, constraint labels are generated for each sample slice. Constraint labels include allowed advancement, restricted advancement, support priority, and verification collection. By concatenating the fixed fields of process step distance variable, cross-section state variable and constraint label in the correct order, process constraint sample and cross-section state vector are obtained.

4. The intelligent optimization method for the step distance in the track CRD underground excavation process as described in claim 3, characterized in that: The construction of the CRD step constraint map based on process constraint samples includes, The four pilot tunnel excavation nodes, temporary intermediate partition wall nodes, initial support nodes, invert arch closure nodes, and secondary lining nodes in the track CRD cut-and-cover section are set as graph nodes. The temporal dependency of adjacent processes, spatial compatibility of the same section, deformation transfer of adjacent sections, closure dependency of support structure, and feedback dependency of monitoring points on construction cycle are set as directed edges. Write the minimum step size, maximum step size, closure waiting time, space occupancy status, deformation transmission direction, and monitoring trigger status into each directed edge, which are extracted from the process constraint samples. When the length of the unclosed support in the cross-section state vector increases or the monitored deformation increment continues to increase, update the directed edge attributes connected to the temporary middle partition wall node, the initial support node and the adjacent pilot tunnel excavation node. When the construction cycle completes the support closure and the monitoring data enters a stable update interval, the restriction marker of the corresponding directed edge is removed. By traversing the graph, the set of preceding processes, the set of mutually exclusive processes, and the set of monitoring constraints for each candidate advancement process are obtained, thus generating the CRD step distance constraint graph.

5. The intelligent optimization method for the step distance in the track CRD underground excavation process as described in claim 4, characterized in that: The calculation of the surrounding rock support coupling constraint includes: Read the monitoring constraint set and support closure dependency edge corresponding to the current candidate advancement process from the CRD step distance constraint diagram, and calculate the deviation of the monitored deformation increment from the historical stable interval, the change of the deformation rate from the current section support age, the deviation of the support strain increment from the same surrounding rock grade sample, and the amplification of the unclosed support length from the candidate advancement step distance. The deviation degree, change degree, support strain deviation degree and amplification degree are adaptively normalized according to the surrounding rock grade, support age and pilot tunnel location in the cross-sectional state vector to generate local constraint components corresponding to candidate propulsion procedures. When the local constraint component and the time-dependent edge in the CRD step distance constraint diagram both point to the same pilot tunnel excavation node, the candidate propulsion process is marked as needing to shorten the step distance. When the local constraint component is only associated with the support closure dependent edge, the candidate propulsion process is marked as support closure priority; All local constraint components are aggregated to the corresponding cross section according to the direction of directed edge propagation to obtain the surrounding rock support coupling constraint quantity.

6. The intelligent optimization method for the step distance in the track CRD underground excavation process as described in claim 5, characterized in that: The method for searching the step distance based on the coupling constraint of the surrounding rock support includes... Read the minimum step distance, maximum step distance, set of preceding processes, and set of mutually exclusive processes for each candidate propulsion process from the CRD step distance constraint diagram to form a candidate set of offset step distances; Based on the coupling constraints of the surrounding rock support, the feasibility screening of the candidate set of deviation distances is performed. Candidate distances that violate the set of preceding procedures are eliminated, candidate distances that trigger the set of mutually exclusive procedures are temporarily stored as conflict candidates, and candidate distances that satisfy the temporal dependency, spatial compatibility and monitoring constraints are retained as feasible candidates. Conflict relaxation calculations are performed on the conflict candidates. The conflict relaxation calculations adjust the range of candidate step distance values ​​in the order of priority of support closure, priority of deformation stability, and priority of symmetrical advancement of the pilot tunnel. The feasible candidates are input into the digital section simulation verification model, and the digital section simulation verification model outputs the deformation prediction, support closure status and space occupancy status for the next construction cycle. The candidate step distances verified by simulation are iteratively sorted to obtain the process step distance scheme.

7. The intelligent optimization method for the step distance in the track CRD underground excavation process as described in claim 6, characterized in that: The generation of step size optimization results and system execution instructions include, The process step distance scheme is converted into a process step distance table for the track CRD cut-and-cover section. The process step distance table includes the pilot tunnel number, start and end mileage, target step distance, support closure sequence, allowable advance mark, verification and acquisition mark, and next acquisition trigger condition. When the target step distance of any guide hole in the process step distance scheme is shortened by the conflict relaxation calculation, write the directed edge number and cross-section state vector field corresponding to the shortening reason into the process step distance table. When the deformation prediction output by the digital cross-section simulation verification model exceeds the stable update interval of the same surrounding rock grade, a verification acquisition command is generated, but no advancement command is generated. When the deformation prediction, support closure state, and space occupancy state all satisfy the edge attributes corresponding to the CRD step distance constraint diagram, the process step distance table is encapsulated into a system execution instruction. The system execution instruction is sent to the data acquisition module, simulation verification module, and field terminal display module, and the instruction version number, generation time, and corresponding section state vector are recorded.

8. A track CRD cut-and-cover excavation process step distance intelligent optimization system, employing the track CRD cut-and-cover excavation process step distance intelligent optimization method as described in any one of claims 1 to 7, characterized in that: It includes a spatiotemporal sample construction module, a constraint graph quantity calculation module, and a step size optimization execution module; The spatiotemporal sample construction module is used to collect spatiotemporal data of the track CRD underground excavation process and generate process constraint samples and cross-sectional state vectors. The constraint diagram quantity calculation module is used to construct a CRD step distance constraint diagram based on the process constraint sample and calculate the coupling constraint quantity of the surrounding rock support. The step distance optimization execution module is used to search for process step distance schemes based on the coupling constraints of the surrounding rock support, and generate step distance optimization results and system execution instructions.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the intelligent optimization method for track CRD tunneling operation step distance as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent optimization method for track CRD tunneling operation step distance as described in any one of claims 1 to 7.

Images