Shield muck geological inversion method based on dynamic slip compensation and timing alignment

Abstract

This application relates to the fields of intelligent construction and industrial data processing technology in underground engineering, and discloses a geological inversion method for shield tunnel slag based on dynamic slip compensation and time-series alignment. The method includes: S10, constructing a multi-source synchronous data stream; S20, extracting fluid dynamic features based on vision; S30, establishing a vision-feedback-driven adaptive solid-liquid slip model and calculating the slip velocity, and calculating the nonlinear lag time of slag particles; S40, constructing the time series of heterogeneous physical quantities; S50, dynamic time-series alignment based on physical constraints; and S60, geological inversion. In this way, a "vision-fluid" cross-domain mapping model is established, transforming microscopic visual parameters (roundness) into macroscopic fluid dynamic parameters (drag factor), effectively solving the problem of lag time calculation deviation caused by neglecting solid-liquid slip in traditional methods, and improving the accuracy of geological inversion.

Description

Technical Field

[0001] This application relates to the fields of intelligent construction of underground engineering and industrial data processing technology, such as a method for geological inversion of shield tunnel slag based on dynamic slip compensation and time alignment. Background Technology

[0002] In large-scale shield tunneling projects such as undersea tunnels and cross-river tunnels, the complex and variable geological environment (such as soft upper layers and hard lower layers, spheroidal weathered bodies, fault fracture zones, etc.) necessitates real-time and accurate understanding of the geological conditions at the tunnel face to ensure construction safety and prevent face instability and seawater intrusion. Currently, obtaining geological information at the tunnel face mainly relies on the analysis of the excavated material discharged from the slurry treatment system.

[0003] However, existing technologies face severe challenges in the slurry discharge conditions of long-distance (typically several kilometers) undersea tunnels. First, manual observation methods suffer from significant time lag, high risk, and strong subjectivity. Second, existing machine vision recognition technologies often focus only on the geometric dimensions of the slurry. Furthermore, existing linear back-calculation methods (i.e., simply dividing the pipe length by the flow velocity to calculate the lag time) cannot address complex spatiotemporal misalignments, leading to a mismatch between the derived geological information and the actual excavation ring number. This can easily mislead adjustments to construction parameters and cause engineering accidents.

[0004] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this application, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] To provide a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. This summary is not intended as a general commentary, nor is it intended to identify key / important components or describe the scope of protection of these embodiments, but rather as a prelude to the detailed description that follows.

[0006] This disclosure provides a method for geological inversion of shield tunnel slag based on dynamic slip compensation and time alignment, in order to improve the accuracy of geological inversion.

[0007] In some embodiments, the shield tunneling slag geological inversion method based on dynamic slip compensation and time-series alignment includes: S10, simultaneously acquiring real-time tunneling parameter data streams of the shield machine and video streams of slag washed at the primary vibrating screen of the slurry treatment plant; wherein, the real-time tunneling parameter data stream includes real-time slurry flow velocity; S20, performing image enhancement and instance segmentation on the slag video stream to extract morphological features of slag particles; wherein, the morphological features include: equivalent particle size and roundness; S30, constructing a solid-liquid two-phase flow slip model, and calculating the specific slip velocity of each slag particle in combination with the equivalent particle size and the roundness; S40: Calculate the nonlinear lag time of each slag particle based on the specific slip velocity and the real-time mud flow rate; S50: Calculate the tunneling specific energy sequence based on the real-time tunneling parameter data stream, and construct a slag geological index sequence based on the morphological characteristics and nonlinear lag time of the slag particles; S60: Introduce physical time window constraints, use a dynamic time warping algorithm to calculate the optimal nonlinear warping path between the tunneling specific energy sequence and the slag geological index sequence, and establish a mapping relationship between the tunneling ring number and slag characteristics; S70: Generate a geological lithology distribution map of the tunnel face based on the mapping relationship, and output adjustment instructions for the shield tunneling parameters.

[0008] In some embodiments, the shield tunneling slag geological inversion system based on dynamic slip compensation and time alignment includes: a data acquisition module for synchronously acquiring real-time tunneling parameter data streams of the shield machine and video streams of slag washed at the primary vibrating screen of the slurry treatment; a memory for storing program instructions; a processor for executing the shield tunneling slag geological inversion method based on dynamic slip compensation and time alignment as described above when running the program instructions; and a display terminal for displaying the geological lithology distribution map of the tunnel face after time alignment correction in real time.

[0009] The shield tunneling slag geological inversion method based on dynamic slip compensation and time-series alignment provided in this disclosure can achieve the following technical effects:

[0010] First, real-time tunneling parameter data streams from the tunnel boring machine and video streams of slag washed at the primary vibrating screen of the slurry treatment system are acquired synchronously. Then, image enhancement and instance segmentation are performed on the slag video streams to extract the morphological features of the slag particles, achieving automated quantification of slag morphology in underground engineering and providing crucial visual foundational data for subsequent fluid dynamics analysis. Second, this method breaks through the limitations of traditional linear estimation of lag time. By constructing a solid-liquid two-phase flow slip model and combining the extracted morphological features, the specific slip velocity of particles and nonlinear lag time are calculated. This mechanism transforms microscopic visual parameters into macroscopic fluid dynamics parameters, physically reconstructing the true motion law of slag in the pipeline and effectively overcoming the lag time calculation deviation problem caused by neglecting the relative slip between solid and liquid in traditional methods. Subsequently, a tunneling specific energy sequence and a slag geological index sequence were constructed, and a physical time window constraint was introduced. The optimal nonlinear programming path was calculated using a dynamic time warping algorithm. This process effectively solved the problem of nonlinear distortion and disorder of the time series caused by the "tomography effect" in long-distance pipelines, achieving precise spatiotemporal alignment of heterogeneous data (mechanical energy and geological feature images) and establishing a reliable mapping relationship between tunneling ring numbers and slag characteristics. Finally, based on this mapping relationship, a geological lithology distribution map of the tunnel face was generated, and adjustment instructions for the shield tunneling parameters were output. Thus, this application constructed a cross-domain mapping model of "vision-fluid dynamics-mechanical response," which not only significantly improved the overall accuracy of geological inversion but also achieved dynamic guidance for shield tunneling parameters to proactively reduce cutterhead wear and improve construction safety and intelligence.

[0011] The above general description and the description below are exemplary and illustrative only and are not intended to limit this application. Attached Figure Description

[0012] One or more embodiments are illustrated by way of example with reference to the accompanying drawings. These illustrations and drawings do not constitute a limitation on the embodiments. Elements having the same reference numerals in the drawings are shown as similar elements. The drawings are not to be scaled. And wherein:

[0013] Figure 1 This is a schematic diagram of a shield tunnel slag geological inversion method based on dynamic slip compensation and time alignment provided in an embodiment of this disclosure;

[0014] Figure 2 This is a schematic diagram of a vision-based fluid dynamics feature extraction method provided in an embodiment of this disclosure;

[0015] Figure 3 This is a schematic diagram of a method for establishing a visual feedback-driven adaptive solid-liquid slip model provided in an embodiment of this disclosure;

[0016] Figure 4This is a schematic diagram illustrating the principle of solid-liquid two-phase flow chromatographic effect and slip model provided in the embodiments of this disclosure;

[0017] Figure 5 This is a schematic diagram of a dynamic timing alignment method based on physical constraints provided in an embodiment of this disclosure;

[0018] Figure 6 This is a comparison and verification chart of the accuracy of geological inversion calculation at the working face provided in the embodiments of this disclosure;

[0019] Figure 7 This is a verification diagram showing the mapping between the morphological characteristics of slag and the specific sliding velocity provided in the embodiments of this disclosure;

[0020] Figure 8 This is a verification diagram of the prediction error and robustness of slag lag time under different slurry discharge conditions provided in the embodiments of this disclosure. Detailed Implementation

[0021] To provide a more detailed understanding of the features and technical content of the embodiments of this disclosure, the implementation of the embodiments of this disclosure will be described in detail below with reference to the accompanying drawings. The accompanying drawings are for illustrative purposes only and are not intended to limit the embodiments of this disclosure. In the following technical description, for ease of explanation, several details are used to provide a full understanding of the disclosed embodiments. However, one or more embodiments may still be implemented without these details. In other cases, well-known structures and devices may be simplified in their depiction to simplify the drawings.

[0022] The terms "first," "second," etc., used in the specification and accompanying drawings of this disclosure are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this disclosure described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion.

[0023] Unless otherwise stated, the term "multiple" means two or more.

[0024] In this embodiment of the disclosure, the character " / " indicates that the objects before and after it are in an "or" relationship. For example, A / B means: A or B.

[0025] The term "and / or" describes an association between objects, indicating that three relationships can exist. For example, A and / or B means: A or B, or A and B.

[0026] The term "correspondence" can refer to an association or binding relationship. The correspondence between A and B means that there is an association or binding relationship between A and B.

[0027] Combination Figure 1As shown, this disclosure provides a method for geological inversion of shield tunnel slag based on dynamic slip compensation and time-series alignment, including:

[0028] S10, synchronously acquire the real-time tunneling parameter data stream of the tunnel boring machine and the video stream of the slag and rock washed at the first-stage vibrating screen of the slurry treatment; among which, the real-time tunneling parameter data stream includes the real-time slurry flow rate.

[0029] S20, perform image enhancement and instance segmentation on the slag video stream to extract the morphological features of slag particles; among which, the morphological features include: equivalent particle size and roundness.

[0030] S30, construct a solid-liquid two-phase flow slip model, and calculate the specific slip velocity of each slag particle by combining the equivalent particle size and roundness; calculate the nonlinear hysteresis time of each slag particle based on the specific slip velocity and the real-time mud flow rate.

[0031] S40 calculates the tunneling specific energy sequence based on the real-time tunneling parameter data stream, and constructs the slag geological index sequence based on the morphological characteristics of slag particles and nonlinear lag time.

[0032] S50 introduces physical time window constraints and uses a dynamic time warping algorithm to calculate the optimal nonlinear warping path between the tunneling specific energy sequence and the slag geological index sequence, establishing a mapping relationship between the tunneling ring number and slag characteristics.

[0033] S60 generates a geological and lithological distribution map of the tunnel face based on the mapping relationship and outputs adjustment instructions for the tunnel boring machine parameters.

[0034] First, S10 is executed to construct a multi-source synchronous data stream: the system acquires tunneling parameter data streams from the tunnel boring machine's main control PLC at a frequency of 1 Hz via the OPC UA protocol, including: total thrust (kN), cutterhead torque (kN·m), tunneling speed (mm / min), and real-time mud flow rate (instantaneous flow rate of the mud pump, m³ / h), etc. Simultaneously, a high-protection-level industrial camera is installed above the primary vibrating screen of the slurry treatment plant to acquire video streams of the slag after high-pressure water washing, with a resolution of no less than 1080P and a frame rate of no less than 30 fps. To ensure a unified time base, both the video acquisition server and the tunnel boring machine PLC perform millisecond-level time synchronization via NTP service.

[0035] Then, S20 is executed, performing vision-based fluid dynamics feature extraction: image enhancement and instance segmentation are performed on the slag video stream to extract the morphological features of slag particles. Combined with... Figure 2 As shown, it specifically includes:

[0036] S21. Due to the harsh environment at the slurry discharge outlet, the dark channel prior algorithm is first used to remove water mist interference in the slag video stream to obtain the processed video.

[0037] S22, Subsequently, an improved Mask R-CNN network is used to segment the slag in the processed video. To address the problem of severe stacking and adhesion of slag on the vibrating screen, this embodiment introduces an edge repulsion function in the non-maximum suppression (NMS) stage of the Mask R-CNN network. The segmentation process specifically includes:

[0038] S221, For two candidate boxes with IoU greater than a preset threshold, calculate the gradient direction of their corresponding mask edges;

[0039] S222: If the gradient directions are opposite and the distance is less than the distance threshold, it indicates that these are two physically contacting independent objects. In this case, the IoU overlap penalty weight is reduced to force the retention of two candidate boxes, thereby effectively separating tightly stacked rubble. Optionally, when the gradient angle is greater than the angle threshold, the gradient directions are opposite. Optionally, the angle threshold is set to 150 degrees.

[0040] S23, After the segmentation is completed, calculate the equivalent particle size of each slag particle. D p and roundness The formula for calculating roundness is as follows: =4πA / P 2 ,in A For the projected area, P The circumference is used. In this embodiment, roundness is not only a geometric parameter, but is also defined as a fluid resistance influence factor: The smaller (sharper) the value, the greater the resistance to fluid flow and the more significant the slip.

[0041] In this way, by improving the Mask R-CNN network, the ability to segment tightly stacked slag was enhanced, and the accuracy of feature extraction was improved.

[0042] Then, execute S30 to establish a visual feedback-driven solid-liquid two-phase flow slip model: construct the solid-liquid two-phase flow slip model and calculate the specific slip velocity of each slag particle based on the equivalent particle size and roundness; calculate the nonlinear hysteresis time of each slag particle based on the specific slip velocity and the real-time mud flow rate. Figure 3 As shown, it specifically includes:

[0043] S31, Construct a slip model for solid-liquid two-phase flow:

[0044] ,

[0045] in, V slip This refers to the specific sliding velocity of the slag and stone. V mud (t)for t Real-time mud flow rate at any given moment; D pipe This refers to the inner diameter of the slurry discharge pipe; α cal and β cal The dynamic calibration coefficients for the current geological section represent the particle size resistance sensitivity and shape resistance sensitivity, respectively. The calibration coefficients are obtained by back regression by comparing the time difference between the abrupt change points of tunneling parameters and the abrupt change points of slag properties. m The rheological index; D p The equivalent particle size of the slag stone particles; The roundness of the slag particles.

[0046] S32, Parameter Initialization and Calibration: In the initial stage of tunneling, the model parameters are initialized using empirical values ​​provided by the CFD fluid simulation database. These model parameters include: the dynamic calibration coefficient for particle size resistance sensitivity in the current geological section. α cal Shape drag sensitivity dynamic calibration coefficient β cal and rheological index m .

[0047] During tunneling, known geological abrupt changes (such as transitioning from soft soil to hard rock) are used as natural tracer points, or simulated cuttings with RFID tags are manually dumped. The actual time difference from cutting to discharge is recorded, and the current dynamic calibration coefficient is calculated using back regression. α cal and β cal .

[0048] S33, Slip velocity calculation: (e.g.) Figure 4 As shown, the mud flows faster in the pipeline, while the slag lags behind due to gravity and resistance. Substituting the equivalent particle size and roundness into the above solid-liquid two-phase flow slip model, the specific slip velocity of each slag particle is calculated. The solid-liquid two-phase flow slip model quantifies the physical laws: the larger the particle size and the smaller the roundness (the more irregular the shape), the higher the slip velocity of the slag relative to the mud. V slip The smaller the value (i.e., the more severe the lag).

[0049] S34, calculate the nonlinear hysteresis time of each slag particle based on the specific slip velocity and real-time mud flow rate, including:

[0050] Calculate the first i Corrected production time of individual slag particles t out,i :

[0051] t out,i = t capture,i - Δt delay ,

[0052] in, t capture,i This is the timestamp of the image of the slag particle captured in the slag video stream. Δt delay The fixed mechanical delay time from when the slag leaves the discharge pipe opening to when it is cleaned by the first-stage vibrating screen and reaches the camera area (it is a constant, obtained through on-site calibration).

[0053] Calculate the first by integration. i Corrected production time of individual slag particles t out,i Corresponding original cutting time t cut,i :

[0054] ,

[0055] in, For the first i The nonlinear hysteresis time of each slag stone; L This is the total length of the slurry discharge pipe; V mud (t) for t Real-time mud flow rate at any given moment; For the first i The roundness of each slag stone particle; For the first i The equivalent particle size of each slag stone particle.

[0056] In this way, by using the model to calculate the corrected production time of each slag particle and pre-correcting its timestamp, the traditional "macroscopic unified pipeline delay" is optimized into "microscopic particle-specific delay". This accurately isolates the differences in fluid resistance caused by different particle sizes and shapes, fundamentally solving the problem of disordered slag stratification caused by long-distance slurry discharge, and providing a high-fidelity data source for subsequent time series alignment.

[0057] Then, S40 is executed to construct the time series of heterogeneous physical quantities: the tunneling specific energy sequence is calculated based on the real-time tunneling parameter data stream, and the slag geological index sequence is constructed based on the morphological characteristics and nonlinear lag time of the slag particles. Specifically, this includes:

[0058] S41, Calculate the tunneling specific energy sequence based on the real-time tunneling parameter data stream. Q ref (Specific Energy, SE) Qref The SE value represents the energy required to break a unit volume of rock; a higher SE value indicates a harder formation. Specifically, it includes:

[0059] Based on the real-time tunneling parameter data stream, the tunneling specific energy per unit volume of rock is calculated according to the principle of mechanical work. SE (t) The formula is:

[0060] ,

[0061] in, SE(t) Characterization t The specific energy of tunneling at any given moment; F(t) This represents the total thrust of the cutterhead. N(t) This refers to the rotational speed of the cutter head; T(t) This refers to the cutter head torque; v(t) For shield tunneling speed; A This represents the cross-sectional area of ​​the cutterhead excavation.

[0062] Based on timestamp t Calculate in sequence SE(t) This forms a tunneling specific energy sequence that reflects the mechanical response characteristics of the rock strata at the tunnel face. Q ref .

[0063] S42, a geological index sequence for slag was constructed based on the morphological characteristics and nonlinear hysteresis time of slag particles. Q obs ,include:

[0064] First, a preliminary lithological pattern is determined. If visual identification does not reveal obvious blocky debris (i.e., it is entirely mud or fine sand), it is determined to be a soft soil / clay stratum, and the time window is directly applied. Q obs Set to 0.

[0065] If slag and stone are identified, then for each unit time window, calculate the weighted hardness index of all slag and stone particles within that window. H :

[0066] ,

[0067] This formula reflects geological logic: in-situ fractured hard rocks often exhibit large grain size and sharp edges (roundness). i The smaller, (1- i The larger the weighted hardness index, the higher the weighted hardness index. H The higher the value, the harder the formation.

[0068] In this way, mechanical energy data and geological image data are transformed into comparable time series using S40. Simultaneously, lithological model prediction and soft soil benchmark filling are incorporated during series construction to ensure the continuity of the time series and improve the algorithm's robustness in complex and interactive strata.

[0069] Then, S50 is executed, based on physical constraints for dynamic time-series alignment: a physical time window constraint is introduced, and a dynamic time warping algorithm is used to calculate the optimal nonlinear warping path between the tunneling specific energy sequence and the slag geological index sequence, establishing a mapping relationship between the tunneling ring number and slag characteristics. Combined with... Figure 5 As shown, it specifically includes:

[0070] S51, for tunneling specific energy sequence Q ref and slag geological index sequence Q obs Z-score standardization was performed to eliminate dimensional differences. The Euclidean distance matrix between the standardized tunneling specific energy value and the slag geological index value was calculated. M ( i , j ).

[0071] S52, based on the total length of the slurry discharge pipeline L (e.g., 2000 meters) and the fluctuation range of mud pump flow rate [ Q min ,Q max (e.g., 500 m³ / h to 800 m³ / h), calculate the theoretical minimum lag time of the slag. T min and theoretical maximum lag time T max .

[0072] S53, to prevent erroneous matching that violates physical causality (e.g., matching slag to a time before cutting), a Sakoe-Chiba Band constraint is introduced. The width of the global path search constraint band is set. W This allows for path adjustment. w k =(i,j) Satisfying physical causal constraints:

[0073] ,

[0074] in, Time(i) For the first i The timestamp of each excavation data point Time(j) For the first j The timestamp of the slag data.

[0075] Optionally, WSet as the theoretical lag time range [ T min , T max The redundancy value is 1.2 times that of the previous value.

[0076] S54, within the global path search constraint zone, uses dynamic programming to find the path with the minimum cumulative distance, which represents the true correspondence between the slag characteristics and the tunneling ring number.

[0077] After correction by S30, most of the linear hysteresis has been eliminated, but local disorder (chroism effect) caused by turbulence within the pipe still exists. S50 is calculated using the DTW algorithm with physical time window constraints. Q ref and Q obs The optimal nonlinear regularization path between them eliminates spatiotemporal misalignment, solves the problem of nonlinear distortion and disorder of time series caused by the "tomography effect" in long-distance pipelines, and achieves accurate alignment of heterogeneous data (mechanical energy and geological images).

[0078] Finally, execute S60, Geological Inversion: Generate a geological lithology distribution map of the tunnel face based on the mapping relationship, and output adjustment instructions for the tunnel boring machine parameters. Specifically, this includes:

[0079] S61, nonlinearly map the time axis of the slag geological index sequence back to the tunneling ring coordinate axis to generate a refined geological lithology distribution map along the tunnel axis;

[0080] S62, when it detects that it is about to enter a hard rock protrusion section (based on aligned historical trend extrapolation or combined with advanced geological prediction), outputs a command to reduce thrust and increase the cutter head speed to reduce abnormal tool wear.

[0081] In this way, a closed loop is completed from accurate geological inversion to shield tunnel feedforward control. By predicting abrupt changes at the tunnel face (such as "uneven hardness" or "isolated boulders") in advance using high-precision geological lithology distribution maps, the machine can achieve proactive adaptive adjustment of parameters, effectively avoiding engineering risks such as tool breakage, abnormal wear, and excavation face instability caused by the operator's subjective and delayed judgment.

[0082] In summary, this embodiment provides a shield tunneling slag geological inversion method based on dynamic slip compensation and time-series alignment. First, it synchronously acquires real-time tunneling parameter data streams from the shield machine and video streams of slag washed at the primary vibrating screen of the slurry treatment plant. Then, it performs image enhancement and instance segmentation on the slag video streams to extract the morphological features of the slag particles, achieving automated quantification of slag morphology in underground engineering and providing crucial visual foundation data for subsequent fluid dynamics analysis. Secondly, it breaks through the limitations of traditional linear estimation of lag time. By constructing a solid-liquid two-phase flow slip model and combining the extracted morphological features, it calculates the particle-specific slip velocity and nonlinear lag time. This mechanism transforms microscopic visual parameters into macroscopic fluid dynamics parameters, physically reconstructing the true motion law of slag in the pipeline and effectively overcoming the lag time calculation deviation problem caused by neglecting the relative slip between solid and liquid in traditional methods. Subsequently, a tunneling specific energy sequence and a slag geological index sequence were constructed, and a physical time window constraint was introduced. The optimal nonlinear programming path was calculated using a dynamic time warping algorithm. This process effectively solved the problem of nonlinear distortion and disorder of the time series caused by the "tomography effect" in long-distance pipelines, achieving precise spatiotemporal alignment of heterogeneous data (mechanical energy and geological feature images) and establishing a reliable mapping relationship between tunneling ring numbers and slag characteristics. Finally, based on this mapping relationship, a geological lithology distribution map of the tunnel face was generated, and adjustment instructions for the shield tunneling parameters were output. Thus, this application constructed a cross-domain mapping model of "vision-fluid dynamics-mechanical response," which not only significantly improved the overall accuracy of geological inversion but also achieved dynamic guidance for shield tunneling parameters to proactively reduce cutterhead wear and improve construction safety and intelligence.

[0083] Comprehensive performance analysis:

[0084] To fully verify the performance of the inversion method proposed in this application, this embodiment compares it with traditional methods in three dimensions: computational accuracy, feature extraction accuracy, and algorithm robustness.

[0085] (1) Positioning accuracy verification: such as Figure 6 As shown, the horizontal axis represents the tunnel boring machine (TBM) ring number, and the vertical axis represents the dimensionless true hardness index characterizing the rock strata. The black solid lines in the figure represent the actual hardness distribution at the tunnel face (hard rock protrusions exist from ring 5 to ring 9). The figure shows that the traditional linear algorithm (red solid dot line) is affected by the "chlear effect" of long-distance pipeline slurry discharge, resulting in severe lag and smoothing distortion at the hardness abrupt change points. However, the algorithm based on solid-liquid slip compensation and dynamic time warping (DTW) provided in this method (blue solid triangle line) achieves a high degree of spatiotemporal alignment and fitting between the inversion curve and the true hardness index curve at the tunnel face. This result demonstrates that this method effectively eliminates the time distortion caused by solid-liquid slip, greatly improving the spatial positioning accuracy of geological inversion.

[0086] (2) Verification of feature extraction accuracy: such as Figure 7 As shown, the horizontal axis represents the visually extracted equivalent particle size of the slag. D p The vertical axis represents the roundness of the slag, and the size of the bubbles directly characterizes the slag-specific slip velocity calculated by the solid-liquid two-phase flow slip model. V slip The distribution trend of the bubbles shows that as the equivalent particle size of the slag increases and the roundness decreases (i.e., the shape becomes sharper and more irregular), the flow resistance it experiences in the mud increases significantly, and the corresponding relative slip velocity also shows a nonlinear and rapid increasing trend. This verification figure fully demonstrates that the "vision-fluid" cross-domain mapping model constructed by this method fully conforms to the objective physical laws of fluid dynamics, proving the scientific validity and necessity of introducing microscopic visual parameters into macroscopic slip compensation.

[0087] (3) Verification of algorithm robustness: such as Figure 8 As shown, the horizontal axis represents three different slurry discharge conditions of the tunnel boring machine's slurry treatment system (low flow rate (500m³ / h)). 3 / h), normal flow rate (650m³) 3 / h), high flow rate (800m³) 3 The vertical axis represents the percentage of prediction error for the slag time of the slag material. The comparison in the figure shows that under complex operating conditions with drastic fluctuations in slurry pump flow rate, the traditional estimation method (red bars) exhibits significant instability, with a maximum prediction error exceeding 20%. In contrast, this system (blue bars), due to the introduction of a physical time window constraint based on flow rate fluctuations and a dynamic adaptive parameter calibration mechanism, consistently maintains a prediction error below 5% under all operating conditions. This result fully verifies that this inversion system possesses strong anti-interference capabilities and algorithm robustness when facing harsh and dynamically changing real shield tunneling slurry discharge environments.

[0088] This disclosure provides a shield tunneling slag geological inversion system based on dynamic slip compensation and time-series alignment, comprising: an acquisition module, a memory, a processor, and a display terminal. The acquisition module is used to synchronously acquire real-time tunneling parameter data streams of the shield machine and video streams of slag washed at the primary vibrating screen of the slurry treatment plant. The memory stores program instructions; the processor is used to execute the shield tunneling slag geological inversion method based on dynamic slip compensation and time-series alignment as described above when running the program instructions. The display terminal is used to display in real-time a geological lithology distribution map of the tunnel face after time-series alignment correction.

[0089] The specific implementation process of this system can be found in the description of the above method embodiments, and will not be repeated here.

[0090] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0091] While the specific embodiments of the present invention have been described above, they are not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A shield muck geological inversion method based on dynamic slip compensation and timing alignment, characterized in that, include: S10, synchronously acquire the real-time tunneling parameter data stream of the tunnel boring machine and the video stream of the slag and rock washed at the first-stage vibrating screen of the slurry treatment; wherein, the real-time tunneling parameter data stream includes the real-time slurry flow rate; S20, perform image enhancement and instance segmentation on the slag video stream to extract the morphological features of the slag particles; wherein, the morphological features include: equivalent particle size and roundness. S30, construct a solid-liquid two-phase flow slip model, and calculate the specific slip velocity of each slag particle based on the equivalent particle size and the roundness; calculate the nonlinear hysteresis time of each slag particle based on the specific slip velocity and the real-time mud flow rate. S40, calculate the tunneling specific energy sequence based on the real-time tunneling parameter data stream, and construct the slag geological index sequence based on the morphological characteristics and nonlinear lag time of the slag particles; S50, introduce physical time window constraints, use dynamic time warping algorithm to calculate the optimal nonlinear warping path between the tunneling specific energy sequence and the slag geological index sequence, and establish the mapping relationship between tunneling ring number and slag characteristics; S60, Generate a geological lithology distribution map of the tunnel face based on the mapping relationship, and output adjustment instructions for the shield tunneling parameters.

2. The shield tunneling slag geological inversion method based on dynamic slip compensation and time-series alignment according to claim 1, characterized in that, S20 includes: The dark channel prior algorithm is used to remove water mist interference from the slag video stream to obtain the processed video. The Mask R-CNN network is used to segment the slag in the processed video; wherein, the non-maximum suppression stage of the Mask R-CNN network introduces an edge repulsion force function; After the segmentation is completed, the equivalent particle size and roundness of each slag particle are calculated.

3. The shield tunneling slag geological inversion method based on dynamic slip compensation and time-series alignment according to claim 2, characterized in that, The process of segmenting slag and gravel in the processed video using a Mask R-CNN network includes: Detect the gradient direction of the mask edge corresponding to the two candidate boxes; When the gradient directions are opposite and the distance is less than the distance threshold, the IoU overlap penalty weight is reduced, and two candidate boxes are retained.

4. The shield tunneling slag geological inversion method based on dynamic slip compensation and time-series alignment according to claim 1, characterized in that, In step S30, the construction of a solid-liquid two-phase flow slip model, and the calculation of the specific slip velocity of each slag particle in conjunction with the equivalent particle size and the roundness, includes: Construct the slip model for the solid-liquid two-phase flow: ， in, V slip This refers to the specific sliding velocity of the slag and stone. V mud (t) for t Real-time mud flow rate at any given moment; D pipe This refers to the inner diameter of the slurry discharge pipe; α cal and β cal These are the dynamic calibration coefficients for the current geological section, representing the sensitivity to particle size drag and the sensitivity to shape drag, respectively. m The rheological index; D p The equivalent particle size of the slag stone particles; The roundness of the slag particles; Substituting the equivalent particle size and the roundness into the solid-liquid two-phase flow slip model, the specific slip velocity of each slag particle is calculated.

5. The shield tunneling slag geological inversion method based on dynamic slip compensation and time-series alignment according to claim 4, characterized in that, In the initial stage of tunneling, the parameters of the solid-liquid two-phase flow slip model are initialized using empirical values ​​from a CFD fluid simulation database. These model parameters include: the dynamic calibration coefficient for particle size resistance sensitivity in the current geological section. α cal Shape drag sensitivity dynamic calibration coefficient β cal and rheological index m ; During the tunneling process, the actual time difference between cutting and discharge of the slag is recorded using known geological sections or artificially placed tracer markers, and the dynamic calibration coefficient is calculated by back regression. α cal and β cal .

6. The shield tunneling slag geological inversion method based on dynamic slip compensation and time-series alignment according to claim 1, characterized in that, In step S30, calculating the nonlinear hysteresis time of each slag particle based on the specific slip velocity and the real-time mud flow rate includes: Calculate the first i Corrected production time of individual slag particles t out,i : t out,i = t capture,i - Δt delay ， in, t capture,i This is the timestamp of the image of the slag particle captured in the slag video stream. Δt delay The fixed mechanical delay time from when the slag leaves the discharge pipe opening to when it is cleaned by the primary vibrating screen and reaches the camera area; Calculate the first by integration. i Corrected production time of individual slag particles t out,i Corresponding original cutting time t cut,i ; ， in, For the first i The nonlinear hysteresis time of each slag stone; L This is the total length of the slurry discharge pipe; V mud (t) for t Real-time mud flow rate at any given moment; For the first i The roundness of each slag stone particle; For the first i The equivalent particle size of each slag stone particle.

7. The shield tunneling slag geological inversion method based on dynamic slip compensation and time-series alignment according to claim 1, characterized in that, The step of calculating the tunneling specific energy sequence based on the real-time tunneling parameter data stream includes: Based on the real-time tunneling parameter data stream, the tunneling specific energy per unit volume of rock is calculated according to the mechanical work principle. SE (t) : ， in, SE(t) Characterization t The specific energy of tunneling at any given moment; F(t) This represents the total thrust of the cutterhead. N(t) The rotational speed of the cutter head; T(t) This refers to the cutter head torque; v(t) For shield tunneling speed; A The cross-sectional area of ​​the cutterhead excavation; Based on timestamp t Calculate in sequence SE(t) This forms the tunneling specific energy sequence that reflects the mechanical response characteristics of the rock strata at the tunnel face. Q ref ; The construction of the slag geological index sequence based on the morphological characteristics and nonlinear hysteresis time of the slag particles includes: If the lithology is identified as soft soil or clay, the slag geological index sequence for that time window is set to 0. If slag is identified by lithology, then for each unit time window, calculate the weighted hardness index of all slag particles within that window. H : ， Among them, roundness i The smaller the value, the more likely the rock is to be a hard rock that fractures in situ; the weighted hardness index... H The higher.

8. The shield tunneling slag geological inversion method based on dynamic slip compensation and time-series alignment according to claim 1, characterized in that, The S50 includes: The tunneling specific energy sequence and the slag geological index sequence are Z-score standardized, and the Euclidean distance matrix between the standardized tunneling specific energy value and the slag geological index value is calculated. Based on the total length of the slurry discharge pipeline and the fluctuation range of the mud pump flow rate, the theoretical minimum lag time of the slag is calculated. T min and theoretical maximum lag time T max ; Set the width of the global path search constraint band so that the adjusted path satisfies the physical causality constraint: ， in, Time(i) For the first i The timestamp of each excavation data point Time(j) For the first j Timestamp of each piece of slag data; Within the global path search constraint zone, the path with the minimum cumulative distance is solved by dynamic programming. This path represents the true correspondence between the characteristics of the slag and the tunneling ring number.

9. The shield tunneling slag geological inversion method based on dynamic slip compensation and time-series alignment according to any one of claims 1 to 7, characterized in that, The S60 includes: The time axis of the slag geological index sequence is nonlinearly mapped back to the tunneling ring number coordinate axis to generate a refined geological lithology distribution map along the tunnel axis. When the system detects that the path is about to enter a hard rock protrusion section, it outputs a command to reduce thrust and increase cutterhead speed.

10. A geological inversion system for shield tunnel slag based on dynamic slip compensation and time-series alignment, characterized in that, include: The acquisition module is used to synchronously acquire the real-time tunneling parameter data stream of the tunnel boring machine and the video stream of the slag and rock washed at the first-stage vibrating screen of the slurry treatment. Memory, which stores program instructions; A processor, configured to execute, when running the program instructions, the shield tunneling slag geological inversion method based on dynamic slip compensation and time alignment as described in any one of claims 1 to 9; The display terminal is used to display the geological and lithological distribution map of the working face in real time after time-series alignment correction.

Images