A Dynamic Adaptive Combined Decision-Making Method for Advanced Geological Prediction of Tunnels

By employing a dynamic adaptive combined decision-making method for advanced geological forecasting of tunnels, and utilizing a geological complexity classification and standardized coding system to dynamically match forecasting methods, the problem of redundant detection in tunnel construction has been solved, achieving refined management and improved accuracy of forecasting resources.

CN123113028APending Publication Date: 2026-11-17CHINA RAILWAY FIRST SURVEY & DESIGN INST GRP +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611251626.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-18
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing methods for advanced geological prediction of tunnels lack a dynamic adjustment mechanism, resulting in excessive redundant detection, wasted resources, and insufficient prediction accuracy.

Method used

A dynamic adaptive combination decision-making method for tunnel advanced geological prediction is adopted. Through geological complexity classification, standardized parameter coding system and geophysical data analysis, prediction method combination is dynamically matched to achieve on-demand configuration.

Benefits of technology

This has enabled refined management of forecasting methods, reduced construction time, improved forecast accuracy and decision-making precision, and reduced redundant detection.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123113028A_ABST
    Figure CN123113028A_ABST
Patent Text Reader

Abstract

This invention discloses a dynamic adaptive combination decision-making method for tunnel advanced geological prediction. Existing tunnel advanced geological prediction methods rely on manual experience, leading to excessive redundant detection. This invention first classifies the geological complexity, constructs a knowledge base of detection features for prediction methods, and establishes a parameter coding system. After each excavation cycle, geological sketching and coding of the tunnel face are performed to identify adverse geological features and their changing trends. Data from implemented seismic wave reflection methods and other geophysical methods are collected, and the detection anomalies of each method are extracted and coded according to the coding system. Simultaneously, the remaining effective prediction length of each method is calculated. Based on the identified adverse geological features and their changing trends, combined with the detection anomalies of each method and the remaining prediction length, a combination of supplementary prediction methods is determined. The determined supplementary prediction methods are implemented until excavation breakthrough. This invention enables on-demand configuration of prediction methods, dynamically combining prediction means while ensuring prediction accuracy, achieving refined management of prediction resources.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of tunnel and underground engineering construction safety control technology, specifically relating to a dynamic adaptive combined decision-making method for tunnel advanced geological forecasting. Background Technology

[0002] Advanced geological prediction for tunnels is a crucial means of ensuring construction safety. Commonly used prediction methods include geological survey, seismic wave reflection, transient electromagnetic methods, ground-penetrating radar, and advanced drilling. Each method has its own applicable conditions and limitations. Seismic wave reflection has a long detection range (100-200m) and is relatively accurate in locating structural interfaces such as faults and fracture zones, but its ability to identify karst and water-rich bodies is insufficient. Transient electromagnetic methods have a prediction range of approximately 60m, are sensitive to water-bearing structures, but are easily affected by electromagnetic interference from metal equipment and have weak identification capabilities for non-water-bearing structures. Ground-penetrating radar has high resolution but a short effective detection range, suitable for short-range (≤30m) fine-grained detection and anomaly verification. Advanced drilling is the most direct and reliable method, but it is time-consuming, labor-intensive, costly, and has the problem of providing only a partial view.

[0003] Common adverse geological features at the tunnel face during construction include fault fracture zones, densely jointed zones, alteration zones, lithological contact zones, karst caves, and water-rich zones. The requirements for prediction methods differ significantly depending on the type of feature. For structural features, seismic wave reflection and ground-penetrating radar methods sensitive to geological structures are required; when the features are more severe, advanced horizontal drilling verification is necessary. For water-rich features, transient electromagnetic methods sensitive to water bodies are required; and in areas with low electromagnetic resistivity, advanced horizontal drilling verification is also necessary.

[0004] However, the selection and combination of existing methods mainly rely on human experience, lacking a mechanism for dynamic adjustment based on the actual geological characteristics of the tunnel face and existing geophysical parameters. In engineering practice, to prevent missed detections, a fixed combination of multiple methods is often adopted, resulting in excessive redundant detection, which not only consumes construction time but also wastes forecasting resources. Summary of the Invention

[0005] To overcome the shortcomings of existing technologies, this invention provides a dynamic adaptive combination decision-making method for tunnel advanced geological forecasting, which enables forecasting methods to be configured on demand, dynamically combining forecasting means while ensuring forecasting accuracy, and achieving refined management of forecasting resources.

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

[0007] A dynamic adaptive combined decision-making method for tunnel advanced geological prediction includes the following steps:

[0008] S1: Classify the geological complexity based on the exploration and design data, construct a knowledge base for the detection characteristics of the prediction method, and establish a standardized parameter coding system;

[0009] S2: After each cycle of excavation is completed, geological sketching and coding of the tunnel face are carried out to identify adverse geological features and their changing trends.

[0010] S3: Collect data from implemented seismic reflection methods and other geophysical exploration methods, extract the detection anomaly characteristics of each method, and encode them according to the standardized parameter coding system; at the same time, calculate the remaining effective prediction length of each prediction method;

[0011] S4: Combining the adverse geological features and changing trends of the tunnel face identified in step S2 with the anomaly features detected by each method and the remaining effective forecast length obtained in step S3, dynamically match and determine the combination of supplementary forecast methods.

[0012] S5: Implement the supplementary forecasting method determined in step S4, then return to step S2, and repeat steps S2 to S5 until the tunnel is excavated and connected.

[0013] Furthermore, step S1 specifically includes:

[0014] S1-1: Geological Complexity Classification and Initial Forecasting Strategy Formulation

[0015] The geological complexity of the entire tunnel is divided into four levels: complex, relatively complex, moderately complex, and simple. An initial forecasting strategy is formulated: the seismic wave reflection method is used as the routine forecasting method for bedrock sections, and the single forecasting distance is determined; the ground-penetrating radar method is used as the routine forecasting method for non-bedrock sections.

[0016] S1-2: Construction of a Knowledge Base for Prediction Method Detection Features

[0017] The detection characteristic parameters and applicable priorities of each forecasting method are established as follows:

[0018] 1) Seismic wave reflection method: The effective detection distance is 60-120m. It has high sensitivity to geological structures, low sensitivity to water bodies, and medium resolution. It is suitable for the detection of structurally unfavorable geological formations.

[0019] 2) Transient electromagnetic method: The effective detection distance is no more than 60m. It has low sensitivity to geological structures, high sensitivity to water-bearing bodies, and medium resolution. It is suitable for detecting water-bearing adverse geological conditions.

[0020] 3) Ground Penetrating Radar (GPR): Effective detection range is no more than 30m. It has high sensitivity to geological structures, moderate sensitivity to water bodies, and high resolution. It is suitable for fine location of local adverse geological bodies.

[0021] 4) Advanced drilling method: The effective detection distance is 30-60m. It is highly sensitive to geological structures and water-bearing bodies, and the detection results have the highest reliability. It is suitable for verifying the detection results of the aforementioned geophysical methods and directly revealing anomaly areas.

[0022] S1-3: Development of a standardized parameter coding system.

[0023] Furthermore, steps S1-3 specifically include:

[0024] Establish a standardized logical coding system suitable for computer language expression. The coding system includes coding of forecasting methods, anomaly types, detection characteristics, stratigraphic lithology, confidence level, development degree, change trend and remaining effective forecast length, forming a unified structured coding rule.

[0025] Furthermore, the standardized parameter coding system is as follows:

[0026] 1) Forecasting method coding: Seismic wave reflection method is coded as 01, ground-penetrating radar method is coded as 02, transient electromagnetic method is coded as 03, advanced drilling method is coded as 04, and geological sketching method is coded as 05;

[0027] 2) Anomaly type coding: Fault fracture zone is coded as 01, dense joint zone is coded as 02, alteration zone is coded as 03, lithological contact zone is coded as 04, water-rich structure is coded as 05, water-filled or mud-filled karst cave is coded as 06, waterless karst cave is coded as 07, no anomaly is coded as 08.

[0028] 3) Detection feature coding: A three-digit code is used, with the hundreds digit indicating the prediction method. 1 represents the seismic wave reflection method, 2 represents the ground-penetrating radar method, 3 represents the transient electromagnetic method, 4 represents the advanced drilling method, and 5 represents the geological sketching method. The last two digits indicate the typical feature number of the anomaly under this method. When not applicable, the code is X00, indicating the prediction method.

[0029] 4) Stratigraphic lithology coding: A two-digit code is used, covering sedimentary rocks, igneous rocks, metamorphic rocks, and tectonic rocks, specifically:

[0030] Sedimentary rocks include: limestone (code 01), sandstone (code 02), mudstone (code 03), shale (code 05), pebble soil (code 06), gravelly soil (code 07), blocky soil (code 08), carbonaceous shale (code 09), mudstone interbedded with sandstone (code 10), sandstone interbedded with mudstone (code 11), sandstone interbedded with conglomerate (code 12), conglomerate (code 13), conglomerate interbedded with mudstone (code 14), conglomerate interbedded with sandstone (code 15), gypsum rock (code 16), slate interbedded with sandstone (code 17), sandstone interbedded with slate (code 18), siliceous limestone (code 19), limestone (code 20), limestone interbedded with slate (code 21), sandstone interbedded with shale (code 22), shale interbedded with sandstone (code 23), and dolomite (code 24).

[0031] Igneous rocks include: granite (code 04), granite porphyry (code 25), diorite (code 26), diorite porphyry (code 27), andesite (code 28), basalt (code 29), and porphyry (code 30).

[0032] Metamorphic rocks include: slate (code 31), phyllite (code 32), quartzite (code 33), marble (code 34), marble interbedded with phyllite (code 35), marble interbedded with gneiss (code 36), marble interbedded with schist (code 37), schist (code 38), schist interbedded with phyllite (code 39), schist interbedded with gneiss (code 40), phyllite interbedded with gneiss (code 41), phyllite interbedded with marble (code 42), phyllite interbedded with schist (code 43), gneiss (code 44), gneiss interbedded with phyllite (code 45), gneiss interbedded with marble (code 46), gneiss interbedded with schist (code 47), schist interbedded with marble (code 48), and amphibolite (code 49).

[0033] The structural rock types include: cataclastic rock (coded 50), mylonite (coded 51), fault gouge (coded 52), and fault breccia (coded 53).

[0034] 5) Confidence level coding: 01 is the code for normal, 02 is the code for preliminary anomaly, and 03 is the code for confirmed anomaly;

[0035] 6) Developmental stage coding: weak development is coded as 01, moderate development as 02, and strong development as 03;

[0036] 7) Change trend coding: No obvious change is coded as 01, increased change is coded as 02, and decreased change is coded as 03;

[0037] 8) Remaining effective forecast length: a value in meters.

[0038] Furthermore, the detection feature encoding specifically includes:

[0039] For fault fracture zones: seismic wave reflection method shows chaotic, sawtooth-shaped reflections and a P-wave velocity decrease exceeding a set threshold, coded as 101; ground-penetrating radar method shows chaotic waveforms and discontinuous phase axes, coded as 201; transient electromagnetic method is not applicable, coded as 300; advanced drilling method shows sudden rapid drilling speed, stuck drill, grout leakage, and brecciated core, coded as 401; geological sketching method shows fractured rock mass, visible brecciated or muddy interlayers, and abrupt changes in attitude, coded as 501.

[0040] For densely jointed zones: seismic wave reflection method shows chaotic reflections and a P-wave velocity decrease exceeding a set threshold, coded as 102; ground-penetrating radar method shows intermittent phase axes and large amplitude variations, coded as 202; transient electromagnetic method is not applicable, coded as 300; advanced drilling method shows increased drilling speed and core fragments or disc-shaped cores, coded as 402; geological sketching method shows dense joints and multiple sets of fractures cutting through the rock mass, coded as 502.

[0041] For alteration zones: Seismic wave reflection method shows strong reflected energy, P-wave velocity decrease exceeding the set threshold, and increased Poisson's ratio, coded as 103; Ground penetrating radar method shows rapid signal attenuation and chaotic high-frequency components, coded as 203; Transient electromagnetic method shows low resistivity anomaly, coded as 301; Advanced drilling method shows steady but slow drilling, core samples are sandy, soily, or muddy, and low core recovery rate, coded as 403; Geological sketching method shows altered rock mass with visible kaolinization or chloritization, and is easily broken by hand, coded as 503.

[0042] For lithological contact zones: seismic wave reflection method shows abrupt changes in P-wave velocity and Poisson's ratio on both sides of the interface, coded as 104; ground-penetrating radar method shows a clear, straight, strongly reflective interface, coded as 204; transient electromagnetic method is not applicable, coded as 300; advanced drilling method shows abrupt changes in drilling speed and alteration of lithology, coded as 404; geological sketching method shows abrupt changes in lithology and a clear interface, coded as 504.

[0043] For water-rich structures: seismic wave reflection method shows a significant increase in Poisson's ratio, coded as 105; ground-penetrating radar method shows strong amplitude low-frequency reflection and rapid signal energy attenuation, coded as 205; transient electromagnetic method shows a significant decrease in resistivity, coded as 302; advanced drilling method shows water gushing or high-pressure jet at the wellhead and moist core, coded as 405; geological sketching method shows water seepage or linear water discharge at the working face, coded as 505.

[0044] For water- or mud-filled karst caves: the seismic wave reflection method shows a significant decrease in P-wave velocity and a significant increase in Poisson's ratio, coded as 106; the ground-penetrating radar method shows an arc-shaped hyperbolic diffraction wave, coded as 206; the transient electromagnetic method shows a significant decrease in resistivity, coded as 303; the advanced drilling method shows a sudden drop in drill bit during drilling, with water or mud gushing out, coded as 406; the geological sketching method shows that solution grooves, solution channels or solution holes are visible at the working face, filled with water or muddy material, coded as 506.

[0045] For waterless karst caves: the seismic wave reflection method shows a decrease in P-wave velocity and no significant change in Poisson's ratio, coded as 107; the ground-penetrating radar method shows an arc-shaped hyperbolic diffraction wave, coded as 207; the transient electromagnetic method shows an increase in resistivity, coded as 304; the advanced drilling method shows a sudden drop in drill bit during drilling and no fluid outflow, coded as 408; the geological sketching method shows that solution grooves, solution channels or solution holes are visible at the working face, but no water or mud filling is seen, coded as 507.

[0046] Furthermore, in step S2, the adverse geological features include fault fracture zones, dense joint zones, alteration zones, lithological contact zones, water-rich structures, water-filled or mud-filled karst caves, and waterless karst caves.

[0047] Furthermore, in step S3, the remaining effective forecast length L is calculated according to the following formula:

[0048] L = d1-d2-d3

[0049] Where L is the remaining effective forecast length, d1 is the effective forecast distance of this detection, d2 is the mileage that has been excavated since this detection, and d3 is the overlap length required for continuous detection.

[0050] Furthermore, in step S4, the combination of supplementary forecasting methods is dynamically matched according to the following decision rules:

[0051] 1) The seismic wave reflection method is the basic breakthrough method, applicable to all working face conditions, and can be continuously applied at a predetermined frequency;

[0052] 2) When the tunnel face has adverse geological features such as fault fracture zones, dense joint zones, alteration zones, lithological contact zones, water-filled or mud-filled karst caves, or waterless karst caves, the ground-penetrating radar method shall be used as a supplementary means; if the remaining effective prediction length of the ground-penetrating radar method is insufficient, the ground-penetrating radar method shall be added.

[0053] 3) When the unfavorable geological features of the tunnel face are water-rich structures, water-filled or mud-filled karst caves, the transient electromagnetic method is used as a supplementary means. If the remaining effective prediction length of the transient electromagnetic method is insufficient, the transient electromagnetic method is added. At the same time, the advanced drilling method is used as a supplementary means. If the remaining effective prediction length of the advanced drilling method is insufficient, the advanced drilling method is added to verify the water-rich situation of the area to be excavated.

[0054] 4) When the above-mentioned structural or karst adverse geological features show an aggravating trend, if the remaining effective prediction length of the advance drilling method is insufficient, the advance drilling method shall be increased.

[0055] 5) When adverse geological features at the tunnel face disappear or significantly weaken to an anomaly-free state in subsequent cycles, the supplementary forecasting methods that have been added will no longer be added after the remaining effective forecast length is insufficient, and the basic forecasting strategy will be automatically reverted.

[0056] The beneficial effects of this invention are:

[0057] 1) This invention achieves standardization and quantification of forecast decision-making. By establishing a standardized parameter coding system, it transforms supplementary method decision-making that relies on human experience into a parameter matching process that can be recognized by computers, thereby reducing subjective bias.

[0058] 2) This invention enables on-demand configuration of forecasting methods. It drives forecasting method decision-making through the identification of adverse geological features at the working face—structural features trigger ground-penetrating radar and advanced drilling, water-producing features trigger transient electromagnetic methods and advanced drilling, and the forecasting scheme is automatically upgraded when the features intensify and automatically reverts to the basic scheme when the features weaken or disappear. This forms a dynamic response closed loop of "feature appearance → method supplementation → feature disappearance → method regression", overcoming the blindness of the traditional "comprehensive forecasting regardless of features".

[0059] 3) This invention eliminates redundant detection and reduces construction time; it uses the remaining effective forecast length of the forecasting method as a criterion for whether to add the method; when the remaining length is sufficient, it utilizes existing results and supplements them in a timely manner when it is insufficient, thus avoiding the duplication of methods that already have effective forecasting capabilities.

[0060] 4) This invention achieves a dual-drive approach of “face features + geophysical exploration features”, which fully utilizes both the adverse geological features revealed by the geological sketch of the face and the anomaly features detected by existing geophysical data. The fusion of these two approaches in decision-making improves the accuracy and reliability of the decision. Attached Figure Description

[0061] Figure 1 This is a flowchart illustrating the operation of the present invention. Detailed Implementation

[0062] The present invention will now be described in detail with reference to specific embodiments.

[0063] This invention provides a dynamic adaptive combination decision-making method for tunnel advanced geological forecasting, which enables the forecasting method to be configured on demand, minimizing construction time occupation while ensuring forecasting accuracy, and achieving refined management of forecasting resources.

[0064] like Figure 1 As shown, the dynamic adaptive combined decision-making method for tunnel advanced geological prediction of the present invention includes the following steps:

[0065] Step S1: Based on the exploration and design data and relevant industry standards, classify the geological complexity, construct a knowledge base for the detection characteristics of the prediction method, and establish a standardized parameter coding system; Step S1 specifically includes:

[0066] S1-1: Geological complexity classification and initial forecasting strategy formulation;

[0067] Based on the survey and design data and in accordance with Appendix B of the "Technical Specification for Advanced Geological Prediction of Railway Tunnels" (Q / CR9217-2015), the geological complexity of the entire tunnel is classified into complex, relatively complex, moderately complex, and simple categories, and an initial prediction strategy is formulated. For bedrock sections, the seismic wave reflection method is used as the standard prediction method, and its single prediction distance is determined. For non-bedrock sections, the ground-penetrating radar method is used as the standard prediction method. The baseline prediction distance for the seismic wave reflection method is: 100–120 m / time for simple and moderately complex sections, 80–100 m / time for relatively complex sections, and 60–80 m / time for complex sections. When conditions overlap, the most unfavorable principle is followed, i.e., the shorter prediction distance is taken.

[0068] S1-2: Construction of a knowledge base for the detection features of forecasting methods; establishing the detection feature parameters and applicable priorities of each forecasting method, as detailed below:

[0069] 1) Seismic wave reflection method: The effective detection distance is 60-120m. It has high sensitivity to geological structures, low sensitivity to water bodies, and medium resolution. It is suitable for the detection of structurally unfavorable geological features such as fault fracture zones, densely jointed zones, and lithological contact zones.

[0070] 2) Transient electromagnetic method: The effective detection distance is no more than 60m. It has low sensitivity to geological structures, high sensitivity to water-bearing bodies, and medium resolution. It is suitable for detecting water-bearing unfavorable geological structures such as water-rich structures, water-filled or mud-filled karst caves.

[0071] 3) Ground Penetrating Radar (GPR): Effective detection range is no more than 30m. It has high sensitivity to geological structures and moderate sensitivity to water-bearing bodies. It has high resolution and is suitable for fine positioning of local adverse geological bodies such as karst and fissures.

[0072] 4) Advanced drilling method: The effective detection distance is 30-60m. It is highly sensitive to geological structures and water-bearing bodies, and the detection results have the highest reliability. It is suitable for verifying the detection results of the aforementioned geophysical methods and directly revealing anomaly areas.

[0073] S1-3: Establish a standardized logical coding system suitable for computer language expression. This system should cover key parameters such as forecasting methods, anomaly types, detection characteristics, stratigraphic lithology, confidence level, development level, trend of change, and remaining effective forecast length, forming unified structured coding rules; specifically as follows:

[0074] A standardized parameter coding system is established, creating a standardized logical coding system suitable for computer language expression. This standardized parameter coding system includes coding for forecasting methods, anomaly types, detection characteristics, stratigraphy and lithology, confidence levels, development levels, trend of change, and remaining effective forecast length, specifically:

[0075] 1) Prediction method code (two digits): Seismic wave reflection method is coded as 01, ground-penetrating radar method is coded as 02, transient electromagnetic method is coded as 03, advanced drilling method is coded as 04, and geological sketching method is coded as 05;

[0076] 2) Anomaly type code (two digits): Fault fracture zone is coded as 01, dense joint zone is coded as 02, alteration zone is coded as 03, lithological contact zone is coded as 04, water-rich structure is coded as 05, water-filled or mud-filled karst cave is coded as 06, waterless karst cave is coded as 07, no anomaly is coded as 08.

[0077] 3) Detection Feature Coding (Three Digits): A three-digit code is used. The hundreds digit indicates the prediction method, where 1 represents the seismic wave reflection method, 2 represents the ground-penetrating radar method, 3 represents the transient electromagnetic method, 4 represents the advanced drilling method, and 5 represents the geological sketching method. The last two digits indicate the typical feature number of the anomaly under this method. If the prediction method is not applicable or does not exist, it is represented by X00, as detailed below:

[0078] For fault fracture zones: seismic wave reflection method shows chaotic, sawtooth-shaped reflections and a P-wave velocity decrease exceeding the set threshold, coded as 101; ground-penetrating radar method shows chaotic waveforms and discontinuous phase axes, coded as 201; transient electromagnetic method is not applicable, coded as 300; advanced drilling method shows sudden increase in drilling speed, stuck drill, grout leakage, and brecciated core, coded as 401; geological sketching method shows fractured rock mass, visible brecciated or muddy interlayers, and abrupt changes in attitude, coded as 501.

[0079] For densely jointed zones: seismic wave reflection method shows chaotic reflections and a P-wave velocity decrease exceeding a set threshold, coded as 102; ground-penetrating radar method shows intermittent phase axes and large amplitude variations, coded as 202; transient electromagnetic method is not applicable, coded as 300; advanced drilling method shows increased drilling speed and core fragments or disc-shaped cores, coded as 402; geological sketching method shows dense joints and multiple sets of fractures cutting through the rock mass, coded as 502.

[0080] For alteration zones: Seismic wave reflection method shows strong reflected energy, P-wave velocity decrease exceeding the set threshold, and increased Poisson's ratio, coded as 103; Ground penetrating radar method shows rapid signal attenuation and chaotic high-frequency components, coded as 203; Transient electromagnetic method shows low resistivity anomaly, coded as 301; Advanced drilling method shows steady but slow drilling, core samples are sandy, soily, or muddy, and low core recovery rate, coded as 403; Geological sketching method shows altered rock mass with visible kaolinization or chloritization, and is easily broken by hand, coded as 503.

[0081] For lithological contact zones: seismic wave reflection method shows abrupt changes in P-wave velocity and Poisson's ratio on both sides of the interface, coded as 104; ground-penetrating radar method shows a clear, straight, strongly reflective interface, coded as 204; transient electromagnetic method is not applicable, coded as 300; advanced drilling method shows abrupt changes in drilling speed and alteration of lithology, coded as 404; geological sketching method shows abrupt changes in lithology and a clear interface, coded as 504.

[0082] For water-rich structures: seismic wave reflection method shows a significant increase in Poisson's ratio, coded as 105; ground-penetrating radar method shows strong amplitude low-frequency reflection and rapid signal energy attenuation, coded as 205; transient electromagnetic method shows a significant decrease in resistivity, coded as 302; advanced drilling method shows water gushing or high-pressure jet at the wellhead and moist core, coded as 405; geological sketching method shows water seepage or linear water discharge at the working face, coded as 505.

[0083] For water- or mud-filled karst caves: the seismic wave reflection method shows a significant decrease in P-wave velocity and a significant increase in Poisson's ratio, coded as 106; the ground-penetrating radar method shows an arc-shaped hyperbolic diffraction wave, coded as 206; the transient electromagnetic method shows a significant decrease in resistivity, coded as 303; the advanced drilling method shows a sudden drop in drill bit during drilling, with water or mud gushing out, coded as 406; the geological sketching method shows that solution grooves, solution channels or solution holes are visible at the working face, filled with water or muddy material, coded as 506.

[0084] For waterless karst caves: the seismic wave reflection method shows a decrease in P-wave velocity and no significant change in Poisson's ratio, coded as 107; the ground-penetrating radar method shows an arc-shaped hyperbolic diffraction wave, coded as 207; the transient electromagnetic method shows an increase in resistivity, coded as 304; the advanced drilling method shows a sudden drop in drill bit during drilling and no fluid outflow, coded as 408; the geological sketching method shows that solution grooves, solution channels or solution holes are visible at the working face, but no water or mud filling is seen, coded as 507.

[0085] 4) Stratigraphic Lithology Code (Two Digits): Covers common sedimentary rocks, igneous rocks, metamorphic rocks, and tectonic rocks in tunnel engineering, as detailed below:

[0086] Sedimentary rocks include: limestone (code 01), sandstone (code 02), mudstone (code 03), shale (code 05), pebble soil (code 06), gravelly soil (code 07), blocky soil (code 08), carbonaceous shale (code 09), mudstone interbedded with sandstone (code 10), sandstone interbedded with mudstone (code 11), sandstone interbedded with conglomerate (code 12), conglomerate (code 13), conglomerate interbedded with mudstone (code 14), conglomerate interbedded with sandstone (code 15), gypsum rock (code 16), slate interbedded with sandstone (code 17), sandstone interbedded with slate (code 18), siliceous limestone (code 19), limestone (code 20), limestone interbedded with slate (code 21), sandstone interbedded with shale (code 22), shale interbedded with sandstone (code 23), and dolomite (code 24).

[0087] Igneous rocks include: granite (code 04), granite porphyry (code 25), diorite (code 26), diorite porphyry (code 27), andesite (code 28), basalt (code 29), and porphyry (code 30).

[0088] Metamorphic rocks include: slate (code 31), phyllite (code 32), quartzite (code 33), marble (code 34), marble interbedded with phyllite (code 35), marble interbedded with gneiss (code 36), marble interbedded with schist (code 37), schist (code 38), schist interbedded with phyllite (code 39), schist interbedded with gneiss (code 40), phyllite interbedded with gneiss (code 41), phyllite interbedded with marble (code 42), phyllite interbedded with schist (code 43), gneiss (code 44), gneiss interbedded with phyllite (code 45), gneiss interbedded with marble (code 46), gneiss interbedded with schist (code 47), schist interbedded with marble (code 48), and amphibolite (code 49).

[0089] The tectonic rock types include: crushed rock (coded 50), mylonite (coded 51), fault gouge (coded 52), and fault breccia (coded 53).

[0090] 5) Confidence level coding (two digits): 01 is the code for normal, 02 is the code for preliminary abnormality, and 03 is the code for confirmed abnormality.

[0091] 6) Developmental level coding (two digits): weak development is coded as 01, moderate development is coded as 02, and strong development is coded as 03;

[0092] 7) Change trend coding (two digits): No obvious change is coded as 01, increased change is coded as 02, and decreased change is coded as 03;

[0093] 8) Remaining effective forecast length: The value in meters is not included in the encoding string and is stored as an independent numerical parameter.

[0094] The complete encoding string format for a single forecast record is as follows: forecast method code (2 bits) + lithology code (2 bits) + anomaly type code (2 bits) + confidence level code (2 bits) + development level code (2 bits) + trend code (2 bits) + seismic wave reflection method feature code (3 bits) + ground-penetrating radar method feature code (3 bits) + transient electromagnetic method feature code (3 bits) + advanced drilling method feature code (3 bits) + geological sketching method feature code (3 bits), totaling 2+2+2+2+2+2+3+3+3+3+3 = 27 bits. The remaining effective forecast length is stored as an independent numerical parameter.

[0095] Step S2: After each excavation cycle, perform geological sketching and coding of the tunnel face to identify adverse geological features and their changing trends. If no adverse geological features are identified in the current cycle and the existing features do not change significantly, it is considered normal and no supplementary forecasting measures need to be activated. When any adverse geological feature appears or the existing features worsen, the forecasting method is triggered to make supplementary decisions.

[0096] Adverse geological features include fault fracture zones, densely jointed zones, alteration zones, lithological contact zones, water-rich structures, water-filled or mud-filled karst caves, and waterless karst caves; the trend of change includes the degree of development of the feature, i.e., weak development, moderate development, and strong development; the trend of change includes no change, aggravation, and mitigation. The anomaly type code, development degree code, and trend of change code for each feature are recorded according to the S1-3 coding system.

[0097] Step S3: Collect the completed seismic reflection data and other geophysical data, extract the detection anomaly characteristics of each method and encode them according to the coding system of Steps S1-3, and at the same time obtain the remaining effective prediction length L of each prediction method; the remaining effective prediction length L is calculated according to the following formula:

[0098] L = d1-d2-d3

[0099] Where L is the remaining effective forecast length, d1 is the effective forecast distance of this detection, d2 is the mileage already excavated since this detection, and d3 is the overlap length required for continuous detection. For continuous detection using seismic wave reflection and transient electromagnetic methods, the overlap should be no less than 10m; for continuous detection using ground-penetrating radar, the overlap should be no less than 5m. When the remaining effective forecast length is insufficient to cover the needs of the next cycle of excavation, it is determined that supplementary work is required.

[0100] Step S4: Based on the adverse geological features and trends of the tunnel face identified in Step S2, and combined with the remaining effective prediction lengths of each method obtained in Step S3, dynamically match and supplement the combination of prediction methods:

[0101] 1) The seismic wave reflection method is the basic breakthrough method, applicable to all working face conditions, and can be continuously applied at a predetermined frequency;

[0102] 2) When the tunnel face has adverse geological features such as fault fracture zones, dense joint zones, alteration zones, lithological contact zones, water-filled or mud-filled karst caves, or waterless karst caves, the ground-penetrating radar method shall be used as a supplementary means; if the remaining effective prediction length of the ground-penetrating radar method is insufficient, the ground-penetrating radar method shall be added.

[0103] 3) When the unfavorable geological features of the tunnel face are water-rich structures, water-filled or mud-filled karst caves, the transient electromagnetic method is used as a supplementary means. If the remaining effective prediction length of the transient electromagnetic method is insufficient, the transient electromagnetic method is added. At the same time, the advanced drilling method is used as a supplementary means. If the remaining effective prediction length of the advanced drilling method is insufficient, the advanced drilling method is added to verify the water-rich situation of the area to be excavated.

[0104] 4) When the above-mentioned structural or karst adverse geological features show an aggravating trend, if the remaining effective prediction length of the advance drilling method is insufficient, the advance drilling method shall be increased.

[0105] 5) When adverse geological features at the tunnel face disappear or significantly weaken to an anomaly-free state in subsequent cycles, the supplementary forecasting methods that have been added will no longer be added after the remaining effective forecast length is insufficient, and the basic forecasting strategy will be automatically reverted.

[0106] Step S5: Implement the supplementary forecasting method determined in Step S4, macroscopically control the results of the seismic wave reflection method, comprehensively analyze the existing geophysical data and the supplementary method data, complement each other, reduce the blind spots of individual forecasts, and form the final forecast conclusion; then return to Step S2, and repeat Step S2 to Step S5 until the tunnel excavation is completed.

[0107] The specific implementation method is as follows:

[0108] The present invention will be further described in detail below through specific embodiments, but the implementation of the present invention is not limited thereto.

[0109] The main tunnel passes through the contact zone between gneiss and granite, and the geological complexity is classified as complex. The seismic wave reflection method is used as the basic method for tunnel construction. In the S1 stage, the baseline prediction distance is determined to be 70m / time. The construction of the detection feature knowledge base and the establishment of a standardized parameter coding system are completed.

[0110] This embodiment fully presents the dynamic decision-making response of the entire process of "the appearance of structural features → the aggravation of structural features → the appearance of hydrological features → the disappearance of features", and each stage step corresponds one by one with the technical solution S1 to S5 of this invention.

[0111] 1) Section 1: Structural features appear, supplemented by ground-penetrating radar method (excavation to mileage DK1120+790)

[0112] S2: The geological sketch of the current working face identifies a lithological contact zone, which is of moderate development and shows no significant change compared to the previous cycle, triggering a supplementary decision for the forecasting method. The record is coded as follows: Anomaly type code 04 (lithological contact zone), development level code 02 (moderate development), change trend code 01 (no significant change), lithology code 44 (gneiss).

[0113] S3: Collect completed geophysical data: The seismic wave reflection method detection feature code is 104 (Vp, Poisson's ratio abruptly changes on both sides of the interface), indicating that there is a structural anomaly ahead; the remaining effective prediction length of the ground-penetrating radar method is L=0m; the remaining length of the transient electromagnetic method is L=5m; the remaining length of the advanced drilling method is L=0m.

[0114] S4: Dynamic matching is performed based on the adverse geological features of the tunnel face as the main criterion. The tunnel face has structural features (lithological contact zone). According to rule (2) of step S4, the remaining L=0m of the ground-penetrating radar method is insufficient to cover the tunneling needs of the next cycle, and the ground-penetrating radar method needs to be added. The structural features are moderately developed and not aggravated, which does not meet the conditions for supplementary advanced drilling. Decision result: Add the ground-penetrating radar method.

[0115] S5: Supplementary detection was conducted using ground-penetrating radar (GPR), with an effective predicted distance of 30m. The GPR detection characteristic code was 204 (clear, straight, and highly reflective interface). Combined with the results from the seismic wave reflection method (104), analysis confirmed the existence of a lithological contact zone ahead, with a clear interface and a width of approximately 12m. Support parameters were adjusted promptly based on the prediction results, and the passage was safe.

[0116] This paragraph's encoded string record:

[0117] The geological sketch record encoding string is: 05-44-04-03-02-01-100-200-300-400-504

[0118] (Forecasting Method 05 Geological Sketch + Lithology 44 Gneiss + Anomaly 04 Lithological Contact Zone + Confidence 03 Confirmed + Development 02 Moderately Developed + Trend 01 No Change + Various Geophysical Features Not Applicable X00 + Sketch Features 504)

[0119] The seismic reflection record encoding string is: 01-44-04-03-02-01-104-200-300-400-500

[0120] (Prediction method 01 Seismic wave reflection method + Lithology 44 Gneiss + Anomaly 04 Lithological contact zone + Confidence level 03 Confirmed + Development 02 Moderately developed + Trend 01 No change + TSP characteristics 104 + Other X00)

[0121] 2) Section Two: Structural features are emphasized, and advance drilling is added (excavation to mileage DK1120+798).

[0122] S2: The current geological sketch of the working face shows a significant increase in the characteristics of the lithological contact zone, with the degree of fragmentation changing from moderate to strong, and localized seepage occurring, triggering a supplementary decision for the forecasting method. The record is coded as follows: Anomaly type code 04 (lithological contact zone), development level code 03 (strongly developed), and change trend code 02 (intensified).

[0123] S3: Collect completed geophysical data: The seismic wave reflection method detection feature code is 104 (Vp, Poisson's ratio interface abrupt change), and the anomaly degree is significantly enhanced compared with the previous one; the remaining effective prediction length of the ground-penetrating radar method is L=17m (30m of detection has been completed, 8m of tunneling has been completed, and 5m of overlap has been completed, L=30-8-5=17m); the remaining length of the transient electromagnetic method is L=0m; the remaining length of the advanced drilling method is L=0m.

[0124] S4: Dynamic matching according to rules. The face has structural features and the features are significantly emphasized. According to rule (2) of step S4: the remaining L=17m of the ground-penetrating radar method is sufficient to cover the tunneling needs of the next cycle and no increase is needed; the emphasis of the features triggers rule (4) of step S4 - the remaining L=0m of the advanced drilling method, which is insufficient and the advanced drilling method needs to be added for verification. Decision result: Add 30m of advanced drilling method.

[0125] S5: Implement advance drilling. Advance drilling revealed a fault fracture zone ahead, approximately 6 meters wide, with a high degree of fracture and containing a small amount of groundwater. The detection characteristic code was 401 (suddenly rapid drilling speed, stuck drill, grout leakage, brecciated core). Comprehensive analysis of various methods, complementing each other, confirmed the accurate location and scale of the adverse geological body. Based on the forecast results, advance grouting reinforcement measures were implemented in advance, ensuring safe passage.

[0126] This paragraph's encoded string record:

[0127] The geological sketch record encoding string is: 05-44-04-03-03-02-100-200-300-400-504

[0128] (Development level changes from 02 to 03, trend changes from 01 to 02)

[0129] The seismic reflection record encoding string is: 01-44-04-03-03-02-104-100-200-300-400

[0130] The coded string for the ground-penetrating radar (GPR) record is: 02-44-04-03-03-02-100-204-200-300-400

[0131] The coded string for the advanced drilling method is: 04-44-04-03-03-02-100-200-300-401-500

[0132] 3) Section 3: Stream-like water flow appears; transient electromagnetic method added (excavation to mileage DK1120+804).

[0133] S2: The current geological sketch of the working face identifies a thread-like water outflow feature with a water pressure of approximately 0.4 MPa, significantly higher than the previous one, triggering a supplementary decision for the forecasting method. The record is coded as follows: Anomaly type code 05 (water-rich structure), development level code 03 (strongly developed), and change trend code 02 (worsening).

[0134] S3: Collect completed geophysical data: Seismic wave reflection method detection feature code is 105 (ν significantly increased); Transient electromagnetic method remaining effective prediction length L=0m; Advanced drilling method remaining L=19m (30m detection completed, 6m excavation, 5m overlap, L=30-6-5=19m); Ground penetrating radar method remaining L=11m.

[0135] S4: The tunnel face has a strip-shaped water outflow feature (water-rich structure). According to rule (3) of step S4, the transient electromagnetic method has a remaining L=0m, which is insufficient. The transient electromagnetic method should be increased. The advanced drilling method has a remaining L=19m, which is sufficient to cover the tunneling needs of the next cycle. No need to increase it. Decision result: Increase the transient electromagnetic method by 60m.

[0136] S5: Supplementary detection was conducted using the transient electromagnetic method. The detection characteristic code was 302 (significant decrease in resistivity), indicating the presence of a water-rich structure ahead. Combined with the results of the advance drilling (a high-pressure water-rich body was found 25m ahead, with a water pressure of approximately 0.5MPa), the location and scale of the water-rich structure were confirmed. Advance curtain grouting measures were promptly implemented, preventing a water inrush accident.

[0137] This paragraph's encoded string record:

[0138] The geological sketch record encoding string is: 05-44-05-03-03-02-100-200-300-400-505

[0139] The seismic reflection record encoding string is: 01-44-05-03-03-02-105-200-300-400-500

[0140] The transient electromagnetic method recording encoding string is: 03-44-05-03-03-02-100-200-302-400-500

[0141] The coded string for the ground-penetrating radar (GPR) record is: 02-44-05-03-03-02-100-205-300-400-500

[0142] 4) Paragraph 4: Feature disappearance, revert to basic scheme (excavation to mileage DK1120+850)

[0143] S2: The geological sketch of the working face shows that the adverse geological features have disappeared, the rock mass has been restored to its original integrity, there is no water seepage or tectonic signs, and the trend of change is mitigation. The record is coded as follows: Anomaly type code 08 (no anomaly), development level code 01 (weak development), and trend of change code 03 (mitigation).

[0144] S3: Obtain the remaining effective forecast length for each method: seismic wave reflection method L=25m, ground-penetrating radar method L=0m, transient electromagnetic method L=0m, advanced drilling method L=0m.

[0145] S4: According to rule (5) in step S4, the adverse geological features of the subsequent tunnel face have disappeared. The supplementary forecasting methods that have been added will no longer be added after the remaining effective forecast length is insufficient, and the basic forecasting strategy will be automatically reverted. Decision result: No supplementary methods will be activated, and the single basic scheme of seismic wave reflection method will be reverted.

[0146] S5: Only conventional earthquake wave reflection forecasting is implemented, with a forecast distance of 70m / time. No supplementary forecasting methods are added, which greatly saves forecasting resources and construction time.

[0147] This paragraph's encoded string record:

[0148] The geological sketch record encoding string is: 05-44-08-01-01-03-100-200-300-400-500

[0149] The seismic reflection record encoding string is: 01-44-08-01-01-03-100-200-300-400-500

[0150] Table 1 Summary of the decision-making process for each paragraph

[0151]

[0152] Table 2. Typical Encoding String Comparison Table for Each Stage

[0153]

[0154] The content of this invention is not limited to the embodiments listed. Any equivalent modifications made by those skilled in the art to the technical solutions of this invention by reading this specification are covered by the claims of this invention.

Claims

1. A dynamic adaptive combined decision-making method for tunnel advanced geological prediction, comprising the following steps: S1: Classify the geological complexity based on the exploration and design data, construct a knowledge base for the detection characteristics of the prediction method, and establish a standardized parameter coding system; S2: After each cycle of excavation is completed, geological sketching and coding of the tunnel face are carried out to identify adverse geological features and their changing trends. S3: Collect data from implemented seismic reflection methods and other geophysical exploration methods, extract the detection anomaly characteristics of each method, and encode them according to the standardized parameter coding system; at the same time, calculate the remaining effective prediction length of each prediction method; S4: Combining the adverse geological features and changing trends of the tunnel face identified in step S2 with the anomaly features detected by each method and the remaining effective forecast length obtained in step S3, dynamically match and determine the combination of supplementary forecast methods. S5: Implement the supplementary forecasting method determined in step S4, then return to step S2, and repeat steps S2 to S5 until the tunnel is excavated and connected.

2. The dynamic adaptive combined decision-making method for tunnel advanced geological prediction according to claim 1, characterized in that: Step S1 specifically involves: S1-1: Geological Complexity Classification and Initial Forecasting Strategy Formulation The geological complexity of the entire tunnel is divided into four levels: complex, relatively complex, moderately complex, and simple. An initial forecasting strategy is formulated: the seismic wave reflection method is used as the routine forecasting method for bedrock sections, and the single forecasting distance is determined; the ground-penetrating radar method is used as the routine forecasting method for non-bedrock sections. S1-2: Construction of a Knowledge Base for Prediction Method Detection Features The detection characteristic parameters and applicable priorities of each forecasting method are established as follows: 1) Seismic wave reflection method: The effective detection distance is 60-120m. It has high sensitivity to geological structures, low sensitivity to water bodies, and medium resolution. It is suitable for the detection of structurally unfavorable geological formations. 2) Transient electromagnetic method: The effective detection distance is no more than 60m. It has low sensitivity to geological structures, high sensitivity to water-bearing bodies, and medium resolution. It is suitable for detecting water-bearing adverse geological conditions. 3) Ground Penetrating Radar (GPR): Effective detection range is no more than 30m. It has high sensitivity to geological structures, moderate sensitivity to water bodies, and high resolution. It is suitable for fine location of local adverse geological bodies. 4) Advanced drilling method: The effective detection distance is 30-60m. It is highly sensitive to geological structures and water-bearing bodies, and the detection results have the highest reliability. It is suitable for verifying the detection results of the aforementioned geophysical methods and directly revealing anomaly areas. S1-3: Development of a standardized parameter coding system.

3. The dynamic adaptive combined decision-making method for tunnel advanced geological prediction according to claim 2, characterized in that: Steps S1-3 specifically refer to: Establish a standardized logical coding system suitable for computer language expression. The coding system includes coding of forecasting methods, anomaly types, detection characteristics, stratigraphic lithology, confidence level, development degree, change trend and remaining effective forecast length, forming a unified structured coding rule.

4. The dynamic adaptive combined decision-making method for tunnel advanced geological prediction according to claim 3, characterized in that: The standardized parameter coding system is as follows: 1) Forecasting method coding: Seismic wave reflection method is coded as 01, ground-penetrating radar method is coded as 02, transient electromagnetic method is coded as 03, advanced drilling method is coded as 04, and geological sketching method is coded as 05; 2) Anomaly type coding: Fault fracture zone is coded as 01, dense joint zone is coded as 02, alteration zone is coded as 03, lithological contact zone is coded as 04, water-rich structure is coded as 05, water-filled or mud-filled karst cave is coded as 06, waterless karst cave is coded as 07, no anomaly is coded as 08. 3) Detection feature coding: A three-digit code is used, with the hundreds digit indicating the prediction method. 1 represents the seismic wave reflection method, 2 represents the ground-penetrating radar method, 3 represents the transient electromagnetic method, 4 represents the advanced drilling method, and 5 represents the geological sketching method. The last two digits indicate the typical feature number of the anomaly under this method. When not applicable, the code is X00, indicating the prediction method. 4) Stratigraphic lithology coding: A two-digit code is used, including sedimentary rocks, igneous rocks, metamorphic rocks, and tectonic rocks, specifically: Sedimentary rocks include: limestone (code 01), sandstone (code 02), mudstone (code 03), shale (code 05), pebble soil (code 06), gravelly soil (code 07), blocky soil (code 08), carbonaceous shale (code 09), mudstone interbedded with sandstone (code 10), sandstone interbedded with mudstone (code 11), sandstone interbedded with conglomerate (code 12), conglomerate (code 13), conglomerate interbedded with mudstone (code 14), conglomerate interbedded with sandstone (code 15), gypsum rock (code 16), slate interbedded with sandstone (code 17), sandstone interbedded with slate (code 18), siliceous limestone (code 19), limestone (code 20), limestone interbedded with slate (code 21), sandstone interbedded with shale (code 22), shale interbedded with sandstone (code 23), and dolomite (code 24). Igneous rocks include: granite (code 04), granite porphyry (code 25), diorite (code 26), diorite porphyry (code 27), andesite (code 28), basalt (code 29), and porphyry (code 30). Metamorphic rocks include: slate (code 31), phyllite (code 32), quartzite (code 33), marble (code 34), marble interbedded with phyllite (code 35), marble interbedded with gneiss (code 36), marble interbedded with schist (code 37), schist (code 38), schist interbedded with phyllite (code 39), schist interbedded with gneiss (code 40), phyllite interbedded with gneiss (code 41), phyllite interbedded with marble (code 42), phyllite interbedded with schist (code 43), gneiss (code 44), gneiss interbedded with phyllite (code 45), gneiss interbedded with marble (code 46), gneiss interbedded with schist (code 47), schist interbedded with marble (code 48), and amphibolite (code 49). The structural rock types include: cataclastic rock (coded 50), mylonite (coded 51), fault gouge (coded 52), and fault breccia (coded 53). 5) Confidence level coding: 01 is the code for normal, 02 is the code for preliminary anomaly, and 03 is the code for confirmed anomaly; 6) Developmental stage coding: weak development is coded as 01, moderate development as 02, and strong development as 03; 7) Change trend coding: No obvious change is coded as 01, increased change is coded as 02, and decreased change is coded as 03; 8) Remaining effective forecast length: a value in meters.

5. The dynamic adaptive combined decision-making method for tunnel advanced geological prediction according to claim 4, characterized in that: The detection feature encoding is specifically as follows: For fault fracture zones: seismic wave reflection method shows chaotic, sawtooth-shaped reflections and a P-wave velocity decrease exceeding a set threshold, coded as 101; ground-penetrating radar method shows chaotic waveforms and discontinuous phase axes, coded as 201; transient electromagnetic method is not applicable, coded as 300; advanced drilling method shows sudden rapid drilling speed, stuck drill, grout leakage, and brecciated core, coded as 401; geological sketching method shows fractured rock mass, visible brecciated or muddy interlayers, and abrupt changes in attitude, coded as 501. For densely jointed zones: seismic wave reflection method shows chaotic reflections and a P-wave velocity decrease exceeding a set threshold, coded as 102; ground-penetrating radar method shows intermittent phase axes and large amplitude variations, coded as 202; transient electromagnetic method is not applicable, coded as 300; advanced drilling method shows increased drilling speed and core fragments or disc-shaped cores, coded as 402; geological sketching method shows dense joints and multiple sets of fractures cutting through the rock mass, coded as 502. For alteration zones: Seismic wave reflection method shows strong reflected energy, P-wave velocity decrease exceeding the set threshold, and increased Poisson's ratio, coded as 103; Ground penetrating radar method shows rapid signal attenuation and chaotic high-frequency components, coded as 203; Transient electromagnetic method shows low resistivity anomaly, coded as 301; Advanced drilling method shows steady but slow drilling, core samples are sandy, soily, or muddy, and low core recovery rate, coded as 403; Geological sketching method shows altered rock mass with visible kaolinization or chloritization, and is easily broken by hand, coded as 503. For lithological contact zones: seismic wave reflection method shows abrupt changes in P-wave velocity and Poisson's ratio on both sides of the interface, coded as 104; ground-penetrating radar method shows a clear, straight, strongly reflective interface, coded as 204; transient electromagnetic method is not applicable, coded as 300; advanced drilling method shows abrupt changes in drilling speed and alteration of lithology, coded as 404; geological sketching method shows abrupt changes in lithology and a clear interface, coded as 504. For water-rich structures: seismic wave reflection method shows a significant increase in Poisson's ratio, coded as 105; ground-penetrating radar method shows strong amplitude low-frequency reflection and rapid signal energy attenuation, coded as 205; transient electromagnetic method shows a significant decrease in resistivity, coded as 302; advanced drilling method shows water gushing or high-pressure jet at the wellhead and moist core, coded as 405; geological sketching method shows water seepage or linear water discharge at the working face, coded as 505. For water- or mud-filled karst caves: the seismic wave reflection method shows a significant decrease in P-wave velocity and a significant increase in Poisson's ratio, coded as 106; the ground-penetrating radar method shows an arc-shaped hyperbolic diffraction wave, coded as 206; the transient electromagnetic method shows a significant decrease in resistivity, coded as 303; the advanced drilling method shows a sudden drop in drill bit during drilling, with water or mud gushing out, coded as 406; the geological sketching method shows that solution grooves, solution channels or solution holes are visible at the working face, filled with water or muddy material, coded as 506. For waterless karst caves: the seismic wave reflection method shows a decrease in P-wave velocity and no significant change in Poisson's ratio, coded as 107; the ground-penetrating radar method shows an arc-shaped hyperbolic diffraction wave, coded as 207; the transient electromagnetic method shows an increase in resistivity, coded as 304; the advanced drilling method shows a sudden drop in drill bit during drilling and no fluid outflow, coded as 408; the geological sketching method shows that solution grooves, solution channels or solution holes are visible at the working face, but no water or mud filling is seen, coded as 507.

6. The dynamic adaptive combined decision-making method for tunnel advanced geological prediction according to claim 1, characterized in that: In step S2, the adverse geological features include fault fracture zones, dense joint zones, alteration zones, lithological contact zones, water-rich structures, water-filled or mud-filled karst caves, and waterless karst caves.

7. The dynamic adaptive combined decision-making method for tunnel advanced geological prediction according to claim 1, characterized in that: In step S3, the remaining effective forecast length L is calculated according to the following formula: L = d1-d2-d3 Where L is the remaining effective forecast length, d1 is the effective forecast distance of this detection, d2 is the mileage that has been excavated since this detection, and d3 is the overlap length required for continuous detection.

8. The dynamic adaptive combined decision-making method for tunnel advanced geological prediction according to claim 1, characterized in that: In step S4, the combination of supplementary forecasting methods is dynamically matched according to the following decision rules: 1) The seismic wave reflection method is the basic breakthrough method, applicable to all working face conditions, and can be continuously applied at a predetermined frequency; 2) When the tunnel face has adverse geological features such as fault fracture zones, dense joint zones, alteration zones, lithological contact zones, water-filled or mud-filled karst caves, or waterless karst caves, the ground-penetrating radar method shall be used as a supplementary means; if the remaining effective prediction length of the ground-penetrating radar method is insufficient, the ground-penetrating radar method shall be added. 3) When the unfavorable geological features of the tunnel face are water-rich structures, water-filled or mud-filled karst caves, the transient electromagnetic method is used as a supplementary means. If the remaining effective prediction length of the transient electromagnetic method is insufficient, the transient electromagnetic method is added. At the same time, the advanced drilling method is used as a supplementary means. If the remaining effective prediction length of the advanced drilling method is insufficient, the advanced drilling method is added to verify the water-rich situation of the area to be excavated. 4) When the above-mentioned structural or karst adverse geological features show an aggravating trend, if the remaining effective prediction length of the advance drilling method is insufficient, the advance drilling method shall be increased. 5) When adverse geological features at the tunnel face disappear or significantly weaken to an anomaly-free state in subsequent cycles, the supplementary forecasting methods that have been added will no longer be added after the remaining effective forecast length is insufficient, and the basic forecasting strategy will be automatically reverted.