Fine treatment decision method of complex stratum disaster source fusion multi-source data

Abstract

The present application belongs to the technical field of tunnel construction geological disaster prevention, and discloses a fine treatment decision method of fusion multi-source data of complex stratum disaster source, comprising the following steps: collecting and fusing multi-source geological data, constructing a three-dimensional geological model; generating disaster source risk dynamic evaluation and treatment scheme; based on the fluid-solid coupling similarity theory, using physical model test to verify the preliminary treatment scheme, determining the optimal treatment scheme; executing the optimal treatment scheme, and dynamically regulating and controlling the treatment process; after treatment, evaluating the treatment effect through post-data, and feeding back the effect data to the three-dimensional geological model and knowledge base, realizing dynamic updating of the model and self-learning of the decision system. The present application adopts the above treatment decision method, solves the problems of traditional method relying on experience, information being one-sided and treatment being extensive, realizes advanced accurate prediction and fine and personalized treatment of complex stratum disaster source, and significantly improves the safety and efficiency of tunnel construction.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of tunnel construction geological disaster prevention, and in particular to a fine treatment decision method for complex stratum disaster sources based on multi-source data fusion. BACKGROUND

[0002] In the process of tunnel construction, disaster sources (such as karst stratum, water-rich broken stratum, etc.) in complex stratum (such as karst stratum, water-rich broken stratum, etc.) are prone to cause water inrush, mud inrush, surrounding rock collapse and other accidents, which seriously threaten the safety of construction and the progress of the project. Traditional disaster source detection methods rely on single geophysical prospecting technology or geological drilling, which has the problems of low detection accuracy and limited coverage. At the same time, the treatment scheme is often based on experience and lacks comprehensive support from multi-source data, making it difficult to achieve fine treatment and leading to poor treatment effect or resource waste.

[0003] In the prior art, although some methods attempt to combine multiple geophysical prospecting techniques for disaster source detection, they do not form a systematic multi-source data fusion mechanism, and cannot effectively integrate geophysical prospecting data, geological data and construction monitoring data. In addition, the combination of model testing and numerical simulation is insufficient, and it is difficult to accurately reflect the dynamic response law of disaster sources in the construction process, resulting in a lack of scientific basis for treatment decision-making.

[0004] Therefore, there is an urgent need for an integrated method that fuses multi-source data and realizes fine detection and treatment decision of disaster sources. SUMMARY

[0005] The purpose of the present application is to provide a fine treatment decision method for complex stratum disaster sources based on multi-source data fusion, which realizes the fine management and control of complex stratum disaster sources through multi-source data acquisition, fusion modeling, model test verification, risk assessment and treatment decision.

[0006] To achieve the above purpose, the present application provides a fine treatment decision method for complex stratum disaster sources based on multi-source data fusion, comprising the following steps:

[0007] Step S1, a plurality of geophysical prospecting methods and geological drilling are combined to collect multi-source geological data of the construction area;

[0008] Step S2, the acquired multi-source geological data is fused at data level, feature level and decision level to construct a three-dimensional geological model reflecting the stratum structure and spatial distribution of disaster sources;

[0009] Step S3, based on the three-dimensional geological model, the response of surrounding rock and disaster sources under construction disturbance is analyzed through numerical simulation, and the disaster risk level is dynamically evaluated; combined with the knowledge base and case base, a preliminary treatment scheme is generated;

[0010] Step S4, based on the fluid-structure coupling similarity theory, the preliminary treatment scheme is verified by physical model test, and the treatment scheme is optimized according to the test results to determine the optimal treatment scheme;

[0011] Step S5, in the construction process, the optimal treatment scheme is executed, and the real-time parameters of the construction machinery and the field monitoring data are used to dynamically control the treatment process;

[0012] Step S6, after treatment, the treatment effect is evaluated through post-data, and the effect data is fed back to the three-dimensional geological model and knowledge base to realize dynamic updating of the model and self-learning of the decision system.

[0013] Preferably, in step S1, through the space-air-ground-hole integrated survey technology, multi-scale and multi-dimensional data acquisition of complex stratum disaster sources is realized, and the specific process is as follows:

[0014] Step S11, in the early stage of the project, the water area transient electromagnetic method or the sea area multi-channel shallow seismic method is used to scan and detect the project area, and the distribution of the karst development section is macroscopically mastered, and the project area is preliminarily divided into different sections of weak, medium and strong karst development;

[0015] Step S12, according to the division result, drill holes are arranged in different sections and corresponding geophysical method combinations are selected, specifically including:

[0016] (1) Karst strong / medium development area: elastic wave cross-hole CT as the core method;

[0017] (2) Weak karst development area: supplemented by tube wave detection method and multi-frequency borehole sonar;

[0018] Step S13, in the process of multi-source data acquisition, control geological drill holes are arranged at all geophysical points, core sampling, logging are carried out, and rock mineral composition analysis and physical and mechanical tests are carried out to obtain the true and in-situ parameters of the rock mass, providing calibration points for geophysical interpretation.

[0019] Preferably, in step S2, the obtained multi-source heterogeneous data is fused and processed to build a three-dimensional geological model, and the specific process is as follows:

[0020] Step S21, data level fusion;

[0021] Align and calibrate the original data or preliminary processing results of different geophysical methods in a unified three-dimensional spatial coordinate system;

[0022] Step S22, feature level fusion;

[0023] Firstly, the characteristic information of disaster source is extracted from various data, including: the low-velocity anomaly and its wave velocity value are extracted from elastic wave CT data; the lithology change surface, RQD index and the actual exposed position of karst cave are extracted from drilling data; the three-dimensional point cloud of karst cave wall is extracted from drilling sonar data;

[0024] Then, the obtained characteristic information is combined into a comprehensive feature vector for jointly describing the same geological unit or disaster source;

[0025] Step S23, decision level fusion and three-dimensional geological model construction;

[0026] Step S231, basic geological model construction: based on drilling data and geological profile, the terrain surface, stratum unit and weathering surface are created by automatic / semi-automatic modeling tool of rock-soil mass;

[0027] Step S232, disaster source model construction: the wave velocity three-dimensional numerical simulation model obtained by elastic wave CT inversion is imported into the geological three-dimensional survey design system; the system automatically delineates the development range of karst cave in three-dimensional space by identifying the wave velocity anomaly threshold; the three-dimensional model of disaster source is constructed by using discrete point building or karst professional modeling tool with the range as a constraint;

[0028] Step S233, model integration: the tunnel design model, basic geological model and disaster source model are integrated to construct a three-dimensional geological model, which intuitively shows the spatial relative position relationship between the tunnel and the disaster source.

[0029] Preferably, in step S3, the dynamic influence of tunnel construction on surrounding rock and disaster source is simulated and predicted in advance by using the constructed three-dimensional geological model through numerical simulation technology, and the specific process is as follows:

[0030] Step S311, simulation model establishment;

[0031] Firstly, the constructed three-dimensional geological model is directly imported into the numerical simulation software; different geological bodies in the model are automatically assigned with corresponding rock-soil mass mechanical parameters obtained by experiment;

[0032] Then, the whole process of shield tunneling is simulated, including: the disturbance of cutter head rotation cutting to the front soil, the friction between shield shell and surrounding rock, the formation of shield tail gap, the pressure and filling effect of synchronous grouting;

[0033] Finally, the fluid-structure coupling analysis process is defined in the model to simulate the interaction between groundwater and rock-soil mass in water-rich karst stratum;

[0034] Step S312, after calculation, the response information of surrounding rock and disaster source in the whole construction process is obtained, and multi-physical field response analysis is carried out, including:

[0035] (1) Stress field evolution: analysis of the scope of the plastic zone of surrounding rock, development law, and stress concentration and release at key positions;

[0036] (2) Displacement field evolution: prediction of surface subsidence trough, tunnel convergence deformation, and advanced displacement of soil in front of the working face;

[0037] (3) Seepage field evolution: simulation of the seepage path and pressure distribution change of confined water in the cave-surrounding rock-tunnel system, and prediction of potential water inrush channels;

[0038] (4) Comprehensive identification: through the analysis of the coupling effect of physical fields, it is identified whether the water-resisting rock mass will be unstable, seepage unstable or overall damaged under construction disturbance.

[0039] Preferably, in step S3, based on the output results of numerical simulation, the disaster risk is quantitatively and hierarchically dynamically evaluated, and the specific process is as follows:

[0040] Establish an evaluation index system with safety factor, plastic zone penetration, displacement threshold, seepage velocity / pressure gradient as the core; use the improved risk matrix method to combine the probability and consequences of disaster; wherein, the calculation method of risk value is:

[0041] ;

[0042] Wherein, represents the probability of disaster occurrence, which is based on the numerical simulation results and determined by multi-index comprehensive judgment; represents the plastic zone penetration rate, i.e. the number of plastic zone units / total number of units; when , ; otherwise, the value is taken in proportion; represents the proportion of displacement of the monitoring point exceeding the critical displacement threshold; represents the proportion of seepage velocity or pressure gradient exceeding the critical value; , and are the weights of each index, and ;

[0043] ;

[0044] Wherein, represents the severity of disaster consequences; represents the severity of personnel disaster consequences; represents the severity of equipment disaster consequences; represents the severity of construction period disaster consequences; represents the severity of environmental disaster consequences;

[0045] ;

[0046] wherein, represents the risk value;

[0047] Set the decision threshold , and divide the risk level as follows:

[0048] When , the risk level is low risk, and normal excavation is carried out;

[0049] When , the risk level is medium risk, and the monitoring is strengthened and the treatment is prepared;

[0050] When , the risk level is high risk, and immediate treatment measures are taken and the treatment scheme is executed;

[0051] When , the risk level is extremely high risk, and the design scheme is optimized;

[0052] According to the characteristics of the disaster source and the risk assessment level, one or more preliminary treatment schemes are matched and recommended from the knowledge base; after generating multiple preliminary treatment schemes, a multi-objective decision method is used for comparison and selection; wherein, the comprehensive scoring function is as follows:

[0053] ;

[0054] wherein, represents the comprehensive score of the th treatment scheme; represents the weight of the th evaluation index; represents the normalized score of the th treatment scheme on the th evaluation index; represents the total number of evaluation indexes;

[0055] Finally, the scheme with the highest comprehensive score is selected as the optimal preliminary treatment scheme;

[0056] Then, the recommended treatment scheme and its parameters are re-substituted into the numerical model for digital twin simulation to predict the effect after treatment; through the cycle of calculation-comparison-optimization, the optimal preliminary treatment scheme is selected from multiple alternative schemes.

[0057] Preferably, in step S4, first, based on the fluid-structure coupling similarity theory, a physical model test is used to verify the preliminary treatment scheme, and the specific process is as follows:

[0058] Step S411, based on the fluid-structure coupling similarity theory, the target mechanical parameters of the ideal similar material of the model test are calculated;

[0059] Step S412, adopt CBCS type fluid-solid coupling similar material, its composition includes: aggregate, binder, adjusting agent and mixing agent;

[0060] Through proportioning test, determine the final proportioning, so that the prepared similar material strictly meets the target mechanical parameter value of ideal similar material, and ensures the similarity of the model and the prototype in mechanics and seepage behavior;

[0061] Then, a physical model test system is constructed to simulate the shield tunneling and treatment process in water-rich karst stratum, specifically including:

[0062] Step S421, adopt a modular steel structure to assemble a three-dimensional visual model box, and multiple layers of visual observation windows are arranged on the side surface; a shield machine excavation entrance is reserved, and the boundary distance from the tunnel diameter is greater than 3 times to eliminate the boundary effect;

[0063] Step S422, establish a composite EPB shield tunneling simulation system, which has the functions of cutter head rotation tunneling, hydraulic jacking, soil pressure bin pressure monitoring and double auger soil removal; the tunneling parameters are set and automatically controlled according to the similarity ratio, and are recorded in real time;

[0064] Step S423, based on the real cave shape determined by the three-dimensional geological model, use 3D printing technology to print a disaster source model similar in geometry to the prototype;

[0065] Step S424, construct a water pressure intelligent regulation and control and multi-element information monitoring system, use a water pressure loading device to connect with the disaster source model, and apply and maintain a constant water pressure similar to the prototype water pressure; a micro soil pressure cell, a seepage pressure gauge, a waterproof strain brick and a grating multi-point displacement meter are embedded inside the model body to monitor the evolution of stress, seepage pressure, strain and displacement in real time; a three-dimensional laser scanner is used to scan the model surface at high frequency and non-contact, and a settlement cloud chart of the entire ground surface during construction is constructed; all tunneling parameters of the shield machine are recorded, the state of soil removal and the process of sudden gushing water are recorded automatically;

[0066] Finally, the implementation, verification and optimization of the treatment scheme, specifically including:

[0067] Step S431, test implementation:

[0068] First, the generated preliminary treatment scheme is executed in the model test; then, the shield machine is started and tunneling is carried out according to the set parameters until the disaster source affected area is passed;

[0069] Step S432, scheme verification:

[0070] Under the premise of applying the treatment scheme, tunneling is carried out; by comparing the monitoring data before and after the treatment, the reliability of the scheme is verified;

[0071] Step S433, scheme optimization and optimal scheme determination:

[0072] Parameter inversion: if the treatment effect does not fully meet the expectation, the deficiency of the treatment scheme is inversely analyzed according to the failure data monitored in the test;

[0073] Iterative optimization: based on the inverse analysis result, the parameters of the treatment scheme are adjusted, and the test verification is re-performed.

[0074] Preferably, in step S5, an intelligent construction execution system is built, and the specific process is as follows:

[0075] Step S511, scheme parameter digitalization issuance:

[0076] The key process parameters of the optimal treatment scheme are digitized and directly issued to the corresponding intelligent construction equipment control system;

[0077] Step S512, intelligent equipment precise execution:

[0078] Intelligent grouting: after receiving the instruction, the grouting equipment automatically grouts according to the preset pressure-flow curve, and records the cumulative grouting amount in real time, ensuring that the slurry is accurately filled to the designed range;

[0079] Shield intelligent tunneling: when the shield machine passes through the disaster source influence area, the control system preferentially adopts the recommended parameters to perform automatic tunneling mode, reducing the uncertainty of human operation;

[0080] Mechanized support installation: using automatic equipment such as a mechanical hand, the support components are installed according to the design position in the three-dimensional model.

[0081] Preferably, a digital sensory system is constructed to comprehensively and real-time perceive the construction environment and structure state, real-time collect shield machine parameters and grouting system data, and construct an excavation face stability index and a grouting fullness index;

[0082] Based on the real-time parameters of the shield machine, an excavation face stability index is constructed As shown below:

[0083] ;

[0084] Wherein, represents the measured soil bin pressure; represents the theoretically calculated static soil pressure; represents the measured cutterhead torque; represents the set torque; represents the actual soil volume; represents the theoretical excavation soil volume; 、 and These are weighting coefficients, reflecting the degree of influence of each parameter on stability;

[0085] The regulatory decision is as follows:

[0086] when At that time, the condition was stable, and tunneling continued according to the original parameters;

[0087] when If the excavation face is at risk of instability, the system will automatically prompt you to increase the pressure in the soil chamber. Or reduce the tunneling speed;

[0088] when At this time, mud cake formation or over-excavation may occur, and the system will automatically prompt you to reduce torque. Or adjust the dosage of soil amendment;

[0089] Based on real-time grouting data, a grouting fullness index is constructed. As shown below:

[0090] ;

[0091] in, This indicates the cumulative actual grouting volume; This represents the theoretical grouting volume calculated based on the shield tail gap; This represents the integral of the actual grouting pressure over time. This represents the integral of the set grouting pressure over time;

[0092] Decision-making rules:

[0093] like If the grouting is complete, it can be continued normally.

[0094] like Then dynamic control will be activated: the system will adjust according to... Automatically increases grouting pressure ;in, The gain coefficient is until... Meets the standards;

[0095] Step S522, Geological and structural state perception;

[0096] Internal sensing: In the rock mass surrounding and ahead of the tunnel, deploy or utilize advanced geological prediction systems to acquire in real time: microseismic / acoustic emission signals to monitor the development of rock mass fractures; deep displacement to monitor stratum slippage; and pore water pressure changes to monitor seepage field disturbances.

[0097] Surface sensing: Inside the tunnel, convergence meters and total stations are used to automatically monitor tunnel convergence deformation; on the surface, surveying robots, GB-InSAR, or 3D laser scanners are used to continuously monitor changes in the surface settlement field.

[0098] Preferably, the real-time sensed data is compared with the prediction model and control threshold to achieve dynamic optimization and adjustment of the construction parameters, specifically including:

[0099] Step S531, data fusion and state diagnosis:

[0100] The obtained multi-element real-time data is fused and analyzed in a unified data platform; by real-time comparison with the safety threshold predicted by numerical simulation, the current construction state is diagnosed;

[0101] Step S532, based on the diagnosis result, the system generates a control instruction or issues a warning to the operator, and gives a control suggestion:

[0102] Shield parameter control: if the diagnosis finds that the front rock mass is weak and deforms greatly, the system automatically or prompts the operator to appropriately reduce the tunneling speed and increase the soil chamber pressure to stabilize the excavation face;

[0103] Treatment parameter strengthening: if the diagnosis finds that the deformation of the grouting reinforcement area is still beyond expectation, the system instructs the intelligent grouting system to dynamically increase the grouting pressure or perform compensatory grouting in the specific area;

[0104] Dynamic reinforcement of support: if the initial support stress or deformation is close to the warning value, the system prompts to strengthen the support parameters in the next cycle;

[0105] Step S533, risk warning and emergency intervention:

[0106] A multi-level warning mechanism is established, when the risk level reaches orange or red after multi-element data fusion, the system not only adjusts the parameters, but also automatically triggers an audible and visual alarm, and sets an emergency shutdown interlock to force manual inspection and decision-making.

[0107] Preferably, in step S6, effect evaluation and model updating, the specific process is as follows:

[0108] Step S61, through the post-data collected after construction, the effect of the executed treatment scheme is objectively and quantitatively evaluated;

[0109] Step S611, post-data collection:

[0110] (1) Direct exposure verification includes: drilling core verification and borehole television / optical imaging;

[0111] (2) Non-destructive testing verification includes: cross-hole CT re-measurement and seismic wave method / surface wave method;

[0112] (3) Long-term stability monitoring: continue to collect tunnel convergence, ground settlement, and support structure stress monitoring data for a period of time to confirm whether the deformation and stress have stabilized and are far below the control value;

[0113] Step S612, establishing an effect evaluation index system based on posterior data, specifically including:

[0114] (1) Filling degree index: based on coring and borehole TV, evaluating the filling percentage of slurry to cavity / crack;

[0115] (2) Strength improvement rate: the ratio of rock mass strength / wave velocity value after treatment to that before treatment;

[0116] (3) Integrity index: the area reduction rate of abnormal area in CT image after treatment or the wave velocity improvement rate;

[0117] (4) Stability index: deformation convergence rate and final stable value after treatment;

[0118] According to the above indexes, the comprehensive effect score of treatment is given a quantitative conclusion of excellent, qualified and unqualified; wherein, the quantitative scoring formula of treatment effect is:

[0119]

[0120] wherein, represents the comprehensive effect score; and respectively represent the elastic wave velocity before and after treatment; and respectively represent the area of low-velocity abnormal area in CT image before and after treatment; represents the final displacement value after treatment; represents the allowable displacement value; , and respectively represent the weights of strength, integrity and stability;

[0121] Step S62, dynamic updating of three-dimensional geological model and numerical analysis model;

[0122] Step S621, updating of geological attribute model;

[0123] Disaster source model updating: updating the disaster source model attribute from untreated to slurry-filled or reinforced after treatment, and correcting the geometric boundary and physical and mechanical parameters according to the verification data;

[0124] Rock mass parameter partition updating: according to the strength improvement rate and the wave velocity improvement rate, the rock mass quality grade and mechanical parameters of the reinforced area are partitioned and corrected in the three-dimensional geological model;

[0125] Step S622, numerical analysis model correction;

[0126] ​The updated three-dimensional geological model is reimported into the numerical simulation software, the constitutive model parameters in the numerical model are parameterized inversion by taking the actual deformation and stress data monitored after treatment as the target, so that the numerical simulation result is consistent with the actual situation;

[0127] Step S63, self-learning and evolution of the knowledge base and the case base;

[0128] Successful treatment experience is converted into system storage data as historical experience data;

[0129] Step S631, a new successful treatment case is created, and the data structure comprises:

[0130] (1) Problem description: geological conditions before treatment, disaster source characteristics, risk assessment level, etc.

[0131] (2) Treatment scheme: all details of the finally adopted optimal treatment scheme;

[0132] (3) Effect data: all post-data collected and effect evaluation conclusion;

[0133] (4) Context information: project name, stratum type, hydrological condition;

[0134] Step S632, knowledge rule optimization and generation;

[0135] Verification and reinforcement of existing rules: if the present successful case is highly matched with a recommended rule in the knowledge base, the confidence of the rule will be improved;

[0136] New rule generation: if a new type of optimized treatment scheme is adopted and excellent effect is achieved, a new treatment rule is automatically or semi-automatically generated;

[0137] Failure lesson record: if the treatment effect is unqualified, the case is also recorded in the database, and the reason is analyzed, which is used to issue a warning when similar situations are encountered in the future, so as to avoid repeating the same mistake.

[0138] Therefore, the fine treatment decision method of the complex stratum disaster source by fusing multi-source data has the following beneficial effects:

[0139] (1) Through multi-source geophysical data fusion and three-dimensional geological modeling, the precise positioning of the disaster source is realized, not only the position can be determined, but also the three-dimensional space form, scale, filling state and spatial relationship with the tunnel can be accurately described, and the defect of single geophysical method information is overcome.

[0140] (2) Using numerical simulation based on real geological model, the response of surrounding rock and disaster source under different construction schemes can be dynamically and quantitatively predicted before construction. This makes the risk assessment of water inrush, collapse and other risks no longer stay in qualitative experience judgment, but based on safety factor, displacement field, plastic zone range and other quantitative indexes, realizing the calculable and predictable risk.

[0141] (3) The scheme generation combines the prediction results of numerical simulation and expert knowledge base, and is verified and optimized through physical model test. This ensures that the treatment scheme is a repeatedly verified, personalized optimal solution, which ensures safety while improving economy.

[0142] The technical solutions of the present application will be further described in detail below by means of the drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0143] Figure 1 is the flow chart of the fine treatment decision method of the fusion multi-source data of the complex stratum disaster source of the present application;

[0144] Figure 2 is the flow chart of the multi-source data acquisition of the present application;

[0145] Figure 3 is the flow chart of the fusion processing of the multi-source data of the present application;

[0146] Figure 4 is the three-dimensional visualization model box schematic diagram of the present application;

[0147] Figure 5 is the cloud chart of the ground settlement caused by shield construction of the present application. DETAILED DESCRIPTION

[0148] The technical solutions of the present application will be further described in detail below by means of the drawings and examples.

[0149] As shown in Figure 1 , the fine treatment decision method of the fusion multi-source data of the complex stratum disaster source of the present application comprises the following steps:

[0150] Step S1, multi-source data acquisition: comprehensive multi-geophysical prospecting methods and geological drilling, collecting multi-source geological data of the construction area.

[0151] Through the "space-air-ground-hole" integrated survey technology, multi-scale, multi-dimensional data acquisition of complex stratum, especially marine karst disaster source, can be realized, as shown in Figure 2 .

[0152] Step S11, in the early stage of the project, the water area transient electromagnetic method or the sea area multi-channel shallow seismic method is used to scan and detect the project area on a large scale, the distribution of the karst development section is macroscopically mastered, and the project area is preliminarily divided into different sections such as "karst micro-development, medium development and strong development".

[0153] Step S12, according to the division result, drill holes are arranged in different sections and corresponding geophysical method combinations are selected, specifically including:

[0154] (1) Karst strong / medium development area: elastic wave cross-hole CT as the core method.

[0155] A PVC pipe with an outer diameter of ≥91mm is lowered into the drill hole, after pulling out the casing, one drill hole is used as a transmitting hole and the other as a receiving hole, an electric spark source is used to excite elastic waves at a preset point distance (1.0m), and a string detector is used to receive signals in the receiving hole. The drill hole spacing is controlled within 25 meters to ensure signal penetration and data quality.

[0156] The elastic wave cross-hole CT method uses the good conductivity of seawater to promote the energy release of the electric spark source, overcomes the shielding effect of electromagnetic methods in the sea area, and can accurately determine the planar distribution of inter-hole karst caves and karst fissures.

[0157] (2) Karst micro-development area: supplemented by tube wave detection method and multi-frequency borehole sonar.

[0158] The tube wave detection method uses a super-magnetic vibration source to generate Stoneley waves in the drill hole, and by analyzing the reflection waves of the wave impedance difference interface around the hole, it can detect the undiscovered karst caves within a radius of 0.6-1.0 meters around the drill hole. The multi-frequency borehole sonar lowers the sonar transducer into the karst cave section and performs 360° horizontal rotation scanning, by emitting sound wave pulses and receiving echo signals, it directly measures and fits the three-dimensional spatial form of the single drill hole revealed karst cave.

[0159] Step S13, during the multi-source data acquisition process, control geological drill holes are arranged at all geophysical points, rock cores are taken and recorded, and rock mineral composition analysis and physical and mechanical tests are carried out to obtain the true and in-situ parameters of the rock-soil mass, providing "calibration points" for geophysical interpretation.

[0160] Step S2, multi-source data fusion: the obtained multi-source geological data is fused at the data level, feature level or decision level to construct a three-dimensional geological model reflecting the spatial distribution of stratum structure and disaster sources, as shown in Figure 3 .

[0161] Step S21, data level fusion.

[0162] The raw data or preliminary processing results of different geophysical methods (such as elastic wave CT and tube wave) are aligned and calibrated in a unified three-dimensional spatial coordinate system.

[0163] Step S22, feature level fusion.

[0164] Firstly, the feature information of the disaster source is extracted from various types of data, including: extracting low-velocity abnormal zones and their wave velocity values from elastic wave CT data; extracting lithology change surfaces, RQD indicators, and actual exposed cave locations from drilling data; and extracting cave cavity wall three-dimensional point clouds from drilling sonar data.

[0165] Then, the obtained feature information is combined into a comprehensive feature vector for jointly describing the same geological unit or disaster source.

[0166] Step S23, decision level fusion and three-dimensional geological model construction.

[0167] Step S231, basic geological model construction: based on drilling data and geological profiles, a terrain surface, stratum unit, weathering surface, and other basic geological models are created through an automatic / semi-automatic modeling tool.

[0168] Step S232, three-dimensional model construction of disaster source (cave): the wave velocity three-dimensional numerical simulation model obtained by elastic wave CT inversion is imported into a geological (rock-soil) three-dimensional survey design system (such as GeoStation). The system automatically delineates the development range of the cave in three-dimensional space by identifying the wave velocity anomaly threshold. Using this range as a constraint, a discrete point building or a cave professional modeling tool is used to construct an irregular-shaped, spatially accurate three-dimensional cave model, which is divided into full filling, half filling, and no filling types.

[0169] Step S233, model integration: the tunnel design model (such as BIM model) is integrated with the basic geological model and the disaster source model to construct a three-dimensional geological model, which intuitively displays the spatial relative position relationship between the tunnel and the disaster source, realizes the comprehensive analysis from macro to micro, from two-dimensional profile to three-dimensional space, and from single data interpretation to mutual confirmation of multi-source information, and provides a reliable and detailed “transparent geological” base map for subsequent risk assessment and treatment decision-making.

[0170] Step S3, disaster source risk dynamic assessment and treatment scheme generation: based on the three-dimensional geological model, the response of the surrounding rock and the disaster source under construction disturbance is analyzed through numerical simulation, and the disaster risk level is dynamically evaluated; combined with the knowledge base and case base, a targeted preliminary treatment scheme is generated.

[0171] Step S31, construction disturbance response analysis based on numerical simulation.

[0172] Using the constructed high-precision three-dimensional geological model, the dynamic influence of tunnel construction (such as shield tunneling) on the surrounding rock and the disaster source is simulated and predicted in advance through numerical simulation technology.

[0173] Step S311, simulation model establishment.

[0174] Firstly, the three-dimensional geological model built in step S2 (including the accurate karst cave morphology, stratigraphic interface, fault, etc.) is directly imported into the numerical simulation software (such as ANSYS, FLAC3D, ABAQUS, etc.). Different geological bodies in the model are automatically assigned with the corresponding rock and soil mechanical parameters (such as elastic modulus, Poisson's ratio, cohesion, internal friction angle, permeability coefficient, etc.) obtained through experiments in step S2.

[0175] Then, the whole process of shield tunneling is accurately simulated, including but not limited to: disturbance of cutterhead rotation cutting to the front soil, friction between shield shell and surrounding rock, formation of shield tail gap, pressure and filling effect of synchronous grouting.

[0176] Finally, fluid-structure coupling analysis steps are defined in the model to simulate the interaction between groundwater and rock-soil in water-rich karst strata, especially the seepage path and pressure change of confined cave water under construction disturbance.

[0177] Step S312, multi-physical field response analysis.

[0178] Through calculation, the key response information of surrounding rock and disaster source during the whole construction process is obtained, including:

[0179] (1) Stress field evolution: analysis of the plastic zone range of surrounding rock, development law, and stress concentration and release at key positions (such as tunnel vault, side wall, working face, rock pillar between cave and tunnel).

[0180] (2) Displacement field evolution: prediction of ground subsidence tank, tunnel convergence deformation, and advanced displacement of soil in front of the working face.

[0181] (3) Seepage field evolution: simulation of the seepage path and pressure distribution change of confined water in the cave-surrounding rock-tunnel system, prediction of potential water inrush channel.

[0182] (4) Comprehensive identification: through the analysis of the coupling effect of the above physical fields, it is identified whether the water-resisting rock mass will be unstable, seepage instability (pipe flow) or overall failure under construction disturbance.

[0183] Step S32, dynamic assessment of disaster risk.

[0184] Based on the output results of numerical simulation, the dynamic assessment of disaster risk is quantified and graded.

[0185] Step S321, establishment of risk evaluation index system.

[0186] Establish the evaluation index with safety factor, plastic zone penetration, displacement mutation threshold, seepage velocity / pressure gradient, etc. as the core.

[0187] Step S322, dynamic risk level division.

[0188] The probability of disaster occurrence (based on the ease of instability in numerical simulation) and the consequences (based on the scale of disaster and the impact on the project) are combined using the improved risk matrix method.

[0189] The calculation method of the risk value is:

[0190] ;

[0191] wherein, represents the probability of disaster occurrence, which is based on the numerical simulation results and determined by multi-index comprehensive judgment; represents the plastic zone penetration rate, i.e. the number of plastic zone units / total number of units; when , ; otherwise, the value is taken in proportion; represents the proportion of monitoring points whose displacement exceeds the critical displacement threshold; represents the proportion of seepage velocity or pressure gradient exceeding the critical value; , and are the weights of each index, and , the weights are determined by the analytic hierarchy process (AHP) or historical data inversion.

[0192] ;

[0193] wherein, represents the severity of disaster consequences; represents the severity of personnel disaster consequences; represents the severity of equipment disaster consequences; represents the severity of construction period disaster consequences; represents the severity of environmental disaster consequences; each index is assigned a grade (such as 1-5 points) according to historical cases and expert experience.

[0194] ;

[0195] wherein, represents the risk value.

[0196] Set the decision threshold to divide the risk level, as follows:

[0197] When , the risk level is low risk, and normal excavation;

[0198] When , the risk level is medium risk, and the monitoring is strengthened and the treatment is prepared;

[0199] When , the risk level is high risk, take immediate remedial measures and implement the treatment plan;

[0200] When , the risk level is extremely high risk, optimize the design scheme.

[0201] This evaluation is "dynamic", which means that as the position of the shield tunneling face changes, the disaster source in the different sections in front will be re-evaluated in real time (or in stages).

[0202] Step S33, intelligently generate a preliminary treatment plan.

[0203] Combine the evaluation results with expert knowledge, and intelligently output a targeted treatment plan.

[0204] Step S331, knowledge base and case base construction.

[0205] Knowledge base: store structured treatment method rules, for example: IF (risk level = "high" and disaster type = "high pressure water-rich fracture zone") THEN (recommended method = "advanced curtain grouting").

[0206] Case base: store a large number of historical engineering cases, including disaster geological conditions, successful treatment plans (such as grouting pressure, slurry ratio, support parameters) and treatment effects.

[0207] Step S332, scheme generation and comparison.

[0208] The system matches and recommends one or more preliminary treatment plans from the knowledge base according to the characteristics (type, size, water pressure, and spatial relationship with the tunnel) of the current disaster source and the risk assessment level. The scheme not only indicates the method (such as "full-face advanced deep-hole grouting"), but also gives the range of key process parameters.

[0209] After generating multiple preliminary treatment plans, a multi-objective decision-making method is used for comparison; among them, the comprehensive scoring function is as follows:

[0210] ;

[0211] Among them, represents the comprehensive score of the th treatment plan; represents the weight of the th evaluation index; represents the normalized score of the th treatment plan on the th evaluation index; represents the total number of evaluation indexes.

[0212] Finally, the comprehensive score is selected The highest scheme is selected as the optimal preliminary treatment scheme.

[0213] Then, the recommended treatment scheme and its parameters are re-substituted into the numerical model for "digital twin" simulation to predict the effect after treatment (such as whether the plastic zone is reduced, whether the displacement is controlled, and whether the safety factor is met). Through the cycle of "calculation-comparison-optimization", the optimal preliminary treatment scheme is selected from multiple alternative schemes.

[0214] Step S4, treatment scheme verification and optimization: based on the fluid-structure coupling similarity theory, the preliminary treatment scheme is verified by physical model test, and the treatment scheme is optimized according to the test results to determine the optimal treatment scheme.

[0215] Step S41, fluid-structure coupling similarity theory and material development: the actual engineering prototype (Prototype, P) is scaled down to a laboratory model (Model, M) according to scientific principles.

[0216] Step S411, based on the fluid-structure coupling similarity theory, the target mechanical parameters (density, compressive strength, elastic modulus, permeability coefficient, etc.) of the ideal similar material required for model test are calculated.

[0217] Step S412, similar material development:

[0218] The CBCS type fluid-structure coupling similar material is used, which is composed of:

[0219] Aggregate: standard sand, calcium carbonate, iron powder;

[0220] Binder: white cement, chlorinated paraffin;

[0221] Adjusting agent: silicone oil;

[0222] Mixing agent: water.

[0223] Through strict proportioning tests (such as uniaxial compression, Brazilian split, triaxial compression, and permeability test), the final proportioning is determined to make the physical and mechanical parameters (density, compressive / tensile strength, cohesion, internal friction angle, elastic modulus, permeability coefficient) of the prepared similar material strictly meet the target values calculated in step S41, thereby ensuring the high similarity of the model and the prototype in mechanics and seepage behavior.

[0224] Step S42, construction of physical model test system to simulate the process of shield tunneling and treatment in water-rich karst strata.

[0225] Step S421, three-dimensional visual model box.

[0226] The size is 3m (length) x 2m (width) x 2m (height), and multiple visual observation windows are arranged on the side surface, as shown in Figure 4 .

[0227] The shield tunneling machine excavation entrance is reserved, and the boundary distance from the tunnel hole diameter is greater than 3 times to eliminate the boundary effect.

[0228] Step S422, composite EPB shield tunneling simulation system:

[0229] The system has the functions of cutter head rotation tunneling, hydraulic jacking, soil pressure bin pressure monitoring, and double auger soil removal.

[0230] Core characteristics: Its tunneling parameters (thrust, torque, speed, and tunneling speed) can be accurately set and automatically controlled according to the similarity ratio, and can be recorded in real time.

[0231] Step S423, real form disaster source simulation:

[0232] Based on the real cave form determined in step S2, a 3D printing technology is used to print a disaster source model similar in geometry to the prototype.

[0233] This method solves the problem that traditional pre-embedding methods cannot form complex and irregularly shaped caves and are difficult to remove.

[0234] Step S424, water pressure intelligent regulation and multi-element information monitoring system:

[0235] Water pressure regulation: Use a water pressure loading device to connect with the disaster source model, which can accurately apply and maintain a constant water pressure similar to the prototype water pressure.

[0236] Multi-element data monitoring:

[0237] Internal physical field: Micro soil pressure cells, osmotic pressure gauges, waterproof strain bricks, and grating multi-point displacement meters are buried at key positions in the model body (such as the tunnel periphery, the front of the tunnel face, and the rock pillar between the cave and the tunnel) to monitor the evolution of stress, osmotic pressure, strain, and displacement in real time.

[0238] Surface subsidence: A three-dimensional laser scanner is used to scan the model surface at high frequency and non-contact, and a subsidence cloud chart of the entire surface during construction is constructed, as shown in Figure 5 .

[0239] Construction parameters: Automatically record all tunneling parameters of the shield tunneling machine.

[0240] Macro phenomenon: Record the state of soil removal (such as changes in water content) and the process of sudden gushing water.

[0241] Step S43, implementation, verification, and optimization of treatment scheme.

[0242] Step S431, test implementation:

[0243] First, according to the preliminary treatment scheme, the model test is accurately implemented.

[0244] Then, the shield machine is started and the tunneling is carried out according to the set parameters until the disaster source influence area is passed.

[0245] Step S432, scheme verification:

[0246] Under the premise of applying the treatment scheme, the tunneling is carried out. By comparing the monitoring data before and after the treatment, the reliability of the scheme is verified.

[0247] Verification standard:

[0248] After treatment, whether the displacement of surrounding rock and ground subsidence is effectively controlled within the safety threshold.

[0249] After treatment, whether the plastic zone of surrounding rock is not damaged through.

[0250] After treatment, whether the osmotic pressure is not sharply increased and no gushing water disaster occurs.

[0251] After treatment, whether the tunneling parameters of the shield machine are stable.

[0252] Step S433, scheme optimization and optimal scheme determination:

[0253] Parameter inversion: if the treatment effect does not completely achieve the expected (for example, there is still a large deformation), according to the "failure" data monitored in the test, the shortcomings of the treatment scheme (such as insufficient grouting pressure, insufficient reinforcement range) are analyzed.

[0254] Iterative optimization: based on the analysis results, adjust the parameters of the treatment scheme (such as increasing the grouting pressure, optimizing the slurry ratio, increasing the support stiffness), and retest and verify.

[0255] Through the iterative process of "test-evaluate-optimize-retest", an optimal treatment scheme and its complete set of process parameters are finally determined, which are proved to be safe, economic and efficient in the model test, greatly improving the reliability and success rate of the treatment scheme, and providing the most direct technical support for actual construction.

[0256] Step S5, intelligent construction and dynamic control: in the construction process, the optimal treatment scheme is implemented, and the real-time parameters of the construction machinery and the on-site monitoring data are used to dynamically control the treatment process.

[0257] Step S51, intelligent construction implementation system.

[0258] The optimal treatment scheme determined in step S4 is accurately executed at the construction site by digital and intelligent construction equipment and systems.

[0259] Step S511, scheme parameter digitalization delivery:

[0260] The key process parameters of the optimal treatment scheme (such as grouting pressure and flow, shield tunneling parameters, support structure installation position and timing, etc.) are digitized and directly delivered to the corresponding intelligent construction equipment control system.

[0261] Step S512, intelligent equipment accurate execution:

[0262] Intelligent grouting: After receiving the instruction, the grouting equipment automatically grouts according to the preset pressure-flow curve, and records the cumulative grouting amount in real time, ensuring that the slurry is accurately filled to the designed range.

[0263] Shield intelligent tunneling: When the shield machine passes through the disaster source affected area, the control system preferentially adopts the recommended parameters from steps S3 / S4 to perform automatic tunneling mode, reducing the uncertainty of human operation.

[0264] Mechanized support installation: Using automatic equipment such as mechanical hands, the designed position in the three-dimensional model is used to quickly and accurately install support components such as steel frames and anchor rods.

[0265] Step S52, real-time sensing of multi-element information throughout the process.

[0266] A "digital sense" system is constructed to comprehensively and real-time sense the construction environment and structure state.

[0267] Step S521, construction machinery parameter sensing: Real-time collection of shield machine parameters and grouting system data, construction of excavation face stability index and grouting fullness index.

[0268] Based on the real-time parameters of the shield machine, the excavation face stability index is constructed As follows:

[0269] ;

[0270] Wherein, represents the measured soil bin pressure; represents the theoretically calculated static soil pressure; represents the measured cutterhead torque; represents the set torque; represents the actual soil volume; represents the theoretical excavation soil volume; 、 and These are weighting coefficients, reflecting the degree of influence of each parameter on stability, and can be obtained by training historical data through machine learning.

[0271] The regulatory decision is as follows:

[0272] when At that time, the condition was stable, and tunneling continued according to the original parameters;

[0273] when If the excavation face is at risk of instability, the system will automatically prompt you to increase the pressure in the soil chamber. Or reduce the tunneling speed;

[0274] when At this time, mud cake formation or over-excavation may occur, and the system will automatically prompt you to reduce torque. Alternatively, adjust the dosage of soil amendment.

[0275] Based on real-time grouting data, a grouting fullness index is constructed. As shown below:

[0276] ;

[0277] in, This indicates the cumulative actual grouting volume; This represents the theoretical grouting volume calculated based on the shield tail gap; It represents the integral of the actual grouting pressure over time (reflecting the effect of continuous pressure). This represents the integral of the set grouting pressure over time.

[0278] Decision-making rules:

[0279] like If the grouting is complete, it can be continued normally.

[0280] like Then dynamic control will be activated: the system will adjust according to... Automatically increases grouting pressure ,in The gain coefficient is until... Meets the standard.

[0281] Step S522, Geological and Structural State Perception (“Digital Neural Network”):

[0282] Internal sensing: In the rock mass surrounding and ahead of the tunnel, deploy or utilize advanced geological prediction systems to acquire in real time: microseismic / acoustic emission signals to monitor the development of rock mass fractures; deep displacement (via fixed inclinometers or optical fibers) to monitor stratum slippage; and pore water pressure changes to monitor seepage field disturbances.

[0283] Surface sensing: In the tunnel, convergence meters and automatic total stations are used to monitor the convergence deformation of the tunnel. On the ground, measurement robots, GB-InSAR or three-dimensional laser scanners are used to continuously monitor the changes in the ground subsidence field.

[0284] Step S53, dynamic regulation based on data fusion.

[0285] Compare the real-time sensed data with the prediction model and control threshold to achieve dynamic optimization and adjustment of construction parameters.

[0286] Step S531, data fusion and state diagnosis:

[0287] Fuse and analyze the multi-element real-time data obtained in step S52 in a unified data platform. Through real-time comparison with the safety threshold predicted by numerical simulation in step S3, the current construction state is diagnosed.

[0288] Step S532, intelligent feedback and parameter regulation:

[0289] Based on the above diagnosis results, the system generates control instructions or issues warnings to the operator, and gives control suggestions.

[0290] Shield parameter regulation: If the diagnosis finds that the front rock mass is weak and deforms greatly, the system can automatically or prompt the operator to appropriately reduce the tunneling speed and increase the soil chamber pressure to stabilize the excavation face.

[0291] Treatment parameter strengthening: If the diagnosis finds that the grouting reinforcement area deformation is still beyond expectation, the system can instruct the intelligent grouting system to dynamically increase the grouting pressure or perform compensatory grouting in the specified area.

[0292] Dynamic reinforcement of support: If the initial support stress or deformation is close to the warning value, the system will prompt to strengthen the support parameters in the next cycle, such as thickening the sprayed concrete and densifying the steel frame spacing.

[0293] Step S533, risk warning and emergency intervention:

[0294] Establish a multi-level warning mechanism. When the risk level reaches "orange" or "red" after multi-element data fusion, the system not only adjusts the parameters, but also automatically triggers sound and light alarms, and can set up an emergency shutdown interlock (such as automatically stopping the shield tunneling), to force manual inspection and decision-making to prevent accidents from expanding.

[0295] Step S54, real-time mapping of the digital twin platform.

[0296] The entire process of steps S52 and S53 is visualized in real time on the digital twin platform. The real-time data collected on site drives the three-dimensional model, enabling managers to visually see the real-time state and interaction of the shield machine, monitoring points, and disaster sources in the virtual space, providing the most intuitive decision support for dynamic regulation.

[0297] Based on the above process, the construction process is no longer a static execution of a predetermined plan, but a "self-adaptive system" that can perceive the environment, diagnose the state, predict risks, and optimize itself, ultimately achieving fine and intelligent dynamic treatment of complex stratum disaster sources.

[0298] Step S6, effect evaluation and model updating: After treatment, the treatment effect is evaluated through post-data, and the effect data is fed back to the three-dimensional geological model and knowledge base to realize dynamic updating of the model and self-learning of the decision system.

[0299] Step S61, multi-dimensional post-data collection and effect quantitative evaluation.

[0300] Through the "post-data" collected after construction, the effect of the treatment scheme executed in step S5 is objectively and quantitatively evaluated.

[0301] Step S611, post-data collection.

[0302] (1) Direct exposure verification includes: drilling core verification and borehole television / optical imaging.

[0303] Drilling core verification: verification drilling is performed at key positions in the grouting reinforcement area or after passing through the disaster source. By analyzing the extracted core, the distribution and filling of the grout vein are observed, and the wave velocity value and uniaxial compressive strength of the core are tested and compared with the data before treatment.

[0304] Borehole television / optical imaging: borehole camera equipment is lowered into the verification hole to visually inspect the integrity of the hole wall, the filling condition of the karst cave, and the shape of the grout solid.

[0305] (2) Non-destructive testing verification includes: cross-hole CT re-measurement and seismic wave method / surface wave method.

[0306] Cross-hole CT re-measurement: elastic wave CT or resistivity CT testing is performed again in the treatment area to generate wave velocity / resistivity cloud charts after treatment. By comparing the cloud charts before and after treatment, it is visually shown whether the low-velocity anomaly area / high-conductivity anomaly area disappears, shrinks, or weakens.

[0307] Seismic wave method / surface wave method: such methods are used on the ground surface above the tunnel or in the hole to assess the overall uniformity and strength improvement of the surrounding rock after treatment.

[0308] (3) Long-term stability monitoring: Continue to collect monitoring data such as tunnel convergence, ground settlement, and support structure stress for a period of time to confirm whether the deformation and stress have stabilized and are far below the control value.

[0309] Step S612, effect quantification evaluation.

[0310] Establish an effect evaluation index system based on post-treatment data, which specifically includes:

[0311] (1) Filling degree index: Based on coring and borehole TV, evaluate the filling percentage of slurry to cavities / cracks.

[0312] (2) Strength improvement rate: The ratio of rock mass strength / wave velocity value after treatment to that before treatment.

[0313] (3) Integrity index: Area reduction rate of abnormal area in CT image after treatment or wave velocity improvement rate.

[0314] (4) Stability index: Deformation convergence rate and final stable value after treatment.

[0315] According to the above indexes, the comprehensive effect score of treatment is given the quantitative conclusion of "excellent, qualified, unqualified". Among them, the treatment effect quantification score formula is:

[0316] ;

[0317] Among them, represents the comprehensive effect score; and respectively represent the elastic wave velocity before and after treatment; and respectively represent the area of low-velocity abnormal area in the CT image before and after treatment; represents the final displacement value after treatment; represents the allowable displacement value; , and respectively represent the weights of strength, integrity, and stability.

[0318] Step S62, dynamic updating of three-dimensional geological model and numerical analysis model.

[0319] Use the post-treatment data as "truth data" to correct the previous model to make it infinitely close to reality.

[0320] Step S621, geological attribute model updating.

[0321] Disaster source model updating: Update the disaster source model attribute from "untreated" to "slurry filling" or "reinforcement" after treatment, and correct its geometric boundary and physical and mechanical parameters according to the verification data (such as CT retest results).

[0322] Rock mass parameter partition update: According to the strength improvement rate and wave velocity improvement rate, the rock mass quality grade and mechanical parameters of the reinforcement area are partition corrected in the three-dimensional geological model.

[0323] Step S622, numerical analysis model correction.

[0324] The updated three-dimensional geological model is reimported into the numerical simulation software, and the constitutive model parameters (such as the mechanical behavior of grouting improvement body) in the numerical model are fine-tuned (parameter inversion) to make the numerical simulation results highly consistent with the actual situation.

[0325] The calibrated model can significantly improve the reliability and accuracy of predicting future construction stage (such as the next interval) risks.

[0326] Step S63, self-learning and evolution of knowledge base and case base.

[0327] Convert each successful treatment experience into system storage data as historical experience data.

[0328] Step S631, case structured into the library.

[0329] Create a new successful treatment case, and its data structure includes:

[0330] (1) Problem description: geological conditions before treatment, characteristics of disaster source, risk assessment level, etc.

[0331] (2) Treatment scheme: all details of the final optimal treatment scheme (construction method, equipment, materials, all key process parameters).

[0332] (3) Effect data: all post-data collected in step S61 and effect evaluation conclusion.

[0333] (4) Context information: project name, stratum type, hydrological conditions, etc.

[0334] Step S632, knowledge rule optimization and generation.

[0335] Verify and strengthen existing rules: if the current successful case matches a recommended rule in the knowledge base, the confidence of the rule will be improved.

[0336] Generate new rules: if an innovative or highly optimized treatment scheme is adopted and excellent results are achieved, the system can automatically or semi-automatically generate a new treatment rule.

[0337] Failure Lessons Record: If the treatment effect is "unsatisfactory", the case will also be recorded in the database and the reasons will be analyzed. This will be used to issue warnings when similar situations are encountered in the future, so as to avoid repeating the same mistakes.

[0338] The confidence score (CF) is dynamically updated as follows:

[0339] ;

[0340] in, and These represent the confidence levels before and after the update, respectively. This indicates the score for the effectiveness of this case study. The normalized value (mapped to between 0 and 1); Indicates the learning rate ( This controls the speed at which confidence is updated.

[0341] Based on this, successful cases will increase the confidence of the rules used, while unsuccessful cases will decrease their confidence, thereby enabling the system to learn and evolve on its own.

[0342] Therefore, the present invention adopts the above-mentioned refined treatment decision-making method for complex geological disaster sources by integrating multi-source data. Through multi-source data acquisition, fusion modeling, model test verification, risk assessment and treatment decision-making, it realizes refined management and control of complex geological disaster sources.

[0343] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A refined treatment decision-making method based on the fusion of multi-source data on hazard sources in complex geological formations, characterized in that, Includes the following steps: Step S1: Integrate multiple geophysical exploration methods and geological drilling to collect multi-source geological data of the construction area; Step S2: Perform data-level, feature-level, and decision-level fusion on the acquired multi-source geological data to construct a three-dimensional geological model that reflects the stratigraphic structure and spatial distribution of disaster sources; Step S3: Based on the three-dimensional geological model, the response of the surrounding rock and the disaster source under construction disturbance is analyzed through numerical simulation, and the disaster risk level is dynamically assessed. By combining the knowledge base and case library, a preliminary treatment plan is generated; Step S4: Based on the fluid-structure interaction similarity theory, a physical model experiment is used to verify the preliminary treatment plan, and the treatment plan is optimized based on the experimental results to determine the optimal treatment plan. The specific process is as follows: Step S41: First, based on the fluid-structure interaction similarity theory, a physical model experiment is used to verify the preliminary treatment plan. The specific process is as follows: Step S411: Based on the fluid-structure interaction similarity theory, calculate the target mechanical parameters of the ideal similar material in the model test; Step S412: Use CBCS type fluid-structure interaction similar material, the composition of which includes: aggregate, binder, modifier and mixing agent; Through proportioning experiments, the final proportions are determined so that the physical and mechanical parameters of the prepared similar materials strictly meet the target mechanical parameter values ​​of the ideal similar materials, ensuring the similarity between the model and the prototype in terms of mechanics and percolation behavior; Step S42: Next, construct a physical model test system to simulate the shield tunneling and treatment process in water-rich karst strata, specifically including: Step S421: Assemble a 3D visualization model box using modular steel structure, with multiple visualization observation windows on the side; reserve an excavation entrance for the tunnel boring machine and ensure that its boundary distance is greater than 3 times the tunnel diameter to eliminate boundary effects; Step S422: Establish a composite EPB shield tunneling simulation system, which has the functions of cutterhead rotation tunneling, hydraulic jacking, earth pressure chamber pressure monitoring, and double auger soil removal; its tunneling parameters are set and automatically controlled according to the similarity ratio, and are collected and recorded in real time. Step S423: Based on the actual karst cave morphology determined by the three-dimensional geological model, use 3D printing technology to print a disaster source model that is geometrically similar to the prototype; Step S424: Construct a water pressure intelligent control and multi-dimensional information monitoring system. Connect the water pressure loading device to the disaster source model to apply and maintain a constant water pressure similar to the prototype water pressure. Embed miniature earth pressure cells, piezometers, waterproof strain bricks, and grating multi-point displacement gauges in the model body to monitor the evolution of stress, seepage pressure, strain, and displacement in real time. Use a 3D laser scanner to perform high-frequency, non-contact scanning on the model surface to construct a settlement cloud map of the entire surface during construction. Automatically record all tunneling parameters of the tunnel boring machine, record the soil excavation status, and record the process of sudden water inrush. Step S43: Finally, the implementation, verification, and optimization of the treatment plan include: Step S431, Test Implementation: First, the generated preliminary treatment plan is implemented in a model test; then, the tunnel boring machine is started and tunneled according to the set parameters until it passes through the disaster source's affected area. Step S432, Solution Verification: Under the premise of implementing the treatment plan, tunneling is carried out; the reliability of the plan is verified by comparing the monitoring data before and after the treatment. Step S433, Scheme Optimization and Determination of Optimal Scheme: Parameter inversion: If the treatment effect does not fully meet expectations, the shortcomings of the treatment plan can be analyzed based on the failure data monitored in the experiment. Iterative optimization: Based on the back analysis results, adjust the parameters of the treatment plan and re-verify the experiment; Step S5: During the construction process, implement the optimal treatment plan and use the real-time parameters of the construction machinery and on-site monitoring data to dynamically control the treatment process. Step S6: After treatment, the treatment effect is evaluated through posterior data, and the effect data is fed back to the three-dimensional geological model and knowledge base to realize the dynamic updating of the model and the self-learning of the decision-making system.

2. The refined treatment decision-making method for complex geological disaster sources by fusing multi-source data as described in claim 1, characterized in that, In step S1, multi-scale and multi-dimensional data acquisition of complex geological hazard sources is achieved through integrated air-space-ground-hole exploration technology. The specific process is as follows: Step S11: In the early stage of the project, the transient electromagnetic method in the water or the multichannel shallow seismic method in the sea is used to conduct a range scan and exploration of the project site to macroscopically understand the distribution of karst development sections and preliminarily divide the project area into different sections of karst micro-development, moderate development and strong development. Step S12: Based on the division results, drill holes are strategically deployed in different sections, and appropriate combinations of geophysical exploration methods are selected, specifically including: (1) Karst areas with strong and moderate development: Elastic wave cross-hole CT as the core method; (2) Karst micro-development zone: supplemented by tube wave detection method and multi-frequency borehole sonar; Step S13: During the multi-source data acquisition process, control geological boreholes are set up at all geophysical exploration points to collect and record rock cores, and to conduct rock and mineral composition analysis and physical and mechanical tests to obtain the real and in-situ parameters of the rock and soil mass, providing calibration points for geophysical interpretation.

3. The refined treatment decision-making method for complex geological disaster sources by fusing multi-source data as described in claim 1, characterized in that, In step S2, the acquired multi-source heterogeneous data are fused to construct a three-dimensional geological model. The specific process is as follows: Step S21: Data-level fusion; The raw data or preliminary processing results of different geophysical exploration methods are aligned and calibrated in a unified three-dimensional spatial coordinate system. Step S22: Feature-level fusion; First, characteristic information of disaster sources is extracted from various types of data, specifically including: extracting low-velocity anomaly zones and their wave velocity values ​​from elastic wave CT data; extracting lithological change surfaces, RQD indices, and the actual locations of exposed karst caves from borehole data; and extracting three-dimensional point clouds of karst cave walls from borehole sonar data. Then, the acquired feature information is combined into a comprehensive feature vector to jointly describe the same geological unit or disaster source; Step S23: Decision-level fusion and 3D geological model construction; Step S231, Basic geological model construction: Based on borehole data and geological profile maps, create basic geological models of topographic surfaces, stratigraphic units, and weathering surfaces using automatic or semi-automatic rock and soil modeling tools; Step S232, Disaster Source Model Construction: Import the three-dimensional numerical simulation analysis model of wave velocity obtained by elastic wave CT inversion into the geological three-dimensional exploration and design system; the system automatically delineates the development range of karst caves in three-dimensional space by identifying wave velocity anomaly thresholds; using this range as a constraint, construct the three-dimensional model of the disaster source using discrete point volume building or karst cave professional modeling tools; Step S233, Model Integration: Integrate the tunnel design model with the basic geological model and the disaster source model to construct a three-dimensional geological model, which intuitively shows the spatial relative position of the tunnel and the disaster source.

4. The refined treatment decision-making method for complex geological disaster sources by fusing multi-source data according to claim 1, characterized in that, In step S3, using the constructed three-dimensional geological model, numerical simulation technology is employed to simulate and predict the dynamic impact of tunnel construction on the surrounding rock and hazard sources. The specific process is as follows: Step S311: Establishing the simulation model; First, the constructed three-dimensional geological model is directly imported into the numerical simulation software; different geological bodies in the model are automatically assigned the corresponding rock and soil mechanical parameters obtained through experiments; Then, the entire shield tunneling process was simulated, including: the disturbance of the soil in front by the rotating cutterhead, the friction between the shield shell and the surrounding rock, the formation of the shield tail void, and the pressure and filling effect of synchronous grouting. Finally, a fluid-structure interaction analysis process is defined in the model to simulate the interaction between groundwater and soil in water-rich karst strata. Step S312: After calculation, obtain the response information of the surrounding rock and disaster source throughout the entire construction process, and perform multi-physics response analysis, specifically including: (1) Stress field evolution: Analyze the range and development law of the plastic zone of the surrounding rock, as well as the stress concentration and release in key parts; (2) Displacement field evolution: predicting surface settlement troughs, tunnel convergence deformation, and advance displacement of soil in front of the tunnel face; (3) Evolution of seepage field: Simulate the seepage path and pressure distribution changes of confined water in the cave-surrounding rock-tunnel system to predict potential water inrush channels; (4) Comprehensive identification: By analyzing the coupling effect of the physical field, it can be identified whether the water-retaining rock mass will become unstable, seepage unstable or completely destroyed under construction disturbance.

5. The refined treatment decision-making method for complex geological disaster sources by fusing multi-source data according to claim 4, characterized in that, In step S3, based on the output results of the numerical simulation, a quantitative and hierarchical dynamic assessment of disaster risk is performed. The specific process is as follows: Evaluation indicators were established with safety factor, plastic zone continuity, displacement mutation threshold, seepage velocity, and pressure gradient as the core; an improved risk matrix method was adopted to combine the probability and consequences of disaster occurrence; the risk value was calculated as follows: ； in, It represents the probability of disaster occurrence, based on numerical simulation results and determined through a comprehensive assessment of multiple indicators; This represents the plastic zone penetration rate, i.e., the number of plastic zone elements / the total number of elements; when hour, Otherwise, the value is taken proportionally. This indicates the proportion of monitoring points whose displacement exceeds the critical displacement threshold; This indicates the proportion of seepage velocity or pressure gradient exceeding a critical value; , and Let be the weight of each indicator, and ; ； in, Indicates the severity of the disaster consequences; Indicates the severity of the disaster's consequences for personnel; Indicates the severity of the consequences of equipment damage; Indicates the severity of the consequences of disasters during the construction period; Indicates the severity of the consequences of an environmental disaster; ； in, Indicates the risk value; Set decision threshold The risk levels are classified as follows: when At that time, the risk level was low, and normal tunneling was underway; when At that time, the risk level was medium, and monitoring was strengthened and preparations were made for treatment. when If the risk level is high, take immediate action and implement the treatment plan. when At that time, the risk level was extremely high, and the design scheme was optimized. Based on the characteristics and risk assessment level of the disaster source, one or more preliminary treatment plans are matched and recommended from the knowledge base; after generating multiple preliminary treatment plans, a multi-objective decision-making method is used for comparison and selection; the comprehensive scoring function is as follows: ； in, Indicates the first The overall score of each treatment plan; Indicates the first The weight of each evaluation indicator; Indicates the first The treatment plan was in the first Normalized scores on each evaluation indicator; Indicates the total number of evaluation indicators; Ultimately, the overall score is selected. The highest-level plan is considered the optimal initial treatment plan. Then, the recommended treatment plan and its parameters are substituted back into the numerical model for digital twin simulation to predict the effect after treatment; through a cycle of calculation-comparison-optimization, the optimal preliminary treatment plan is selected from multiple alternative plans.

6. The refined treatment decision-making method for complex geological disaster sources by fusing multi-source data according to claim 1, characterized in that, In step S5, an intelligent construction execution system is established, and the specific process is as follows: Step S511: Digital distribution of solution parameters: The key process parameters of the optimal treatment plan are digitized and directly sent to the corresponding intelligent construction equipment control system. Step S512, Intelligent equipment executes precisely: Intelligent grouting: After receiving the instruction, the grouting equipment automatically performs grouting according to the preset pressure-flow curve and records the cumulative grouting volume in real time to ensure that the grout is accurately filled into the design range; Intelligent tunneling: When the tunnel boring machine passes through the disaster-affected area, the control system prioritizes the use of recommended parameters for automatic tunneling mode, reducing the uncertainty of human operation; Mechanized support installation: Using robotic automated equipment, support components are installed according to the design positions in the three-dimensional model.

7. The refined treatment decision-making method for complex geological disaster sources by fusing multi-source data according to claim 6, characterized in that, A digital sensory system is constructed to perceive the construction environment and structural status in a comprehensive and real-time manner, collect shield machine parameters and grouting system data in real time, and construct the excavation face stability index and grouting fullness index. Based on real-time parameters of the tunnel boring machine, an excavation face stability index is constructed. As shown below: ； in, This indicates the measured pressure in the earth chamber; This represents the theoretically calculated earth pressure at rest. This indicates the measured torque of the cutter head; Indicates the set torque; Indicates the actual amount of excavated material; Indicates the theoretical excavation volume; , and These are weighting coefficients, reflecting the degree of influence of each parameter on stability; The regulatory decision is as follows: when At that time, the condition was stable, and tunneling continued according to the original parameters; when If the excavation face is at risk of instability, the system will automatically prompt you to increase the pressure in the soil chamber. Or reduce the tunneling speed; when At this time, mud cake formation or over-excavation may occur, and the system will automatically prompt you to reduce torque. Or adjust the dosage of soil amendment; Based on real-time grouting data, a grouting fullness index is constructed. As shown below: ； in, This indicates the cumulative actual grouting volume; This represents the theoretical grouting volume calculated based on the shield tail gap; This represents the integral of the actual grouting pressure over time. This represents the integral of the set grouting pressure over time; Decision-making rules: like If the grouting is complete, it can be continued normally. like Then dynamic control will be activated: the system will adjust according to... Automatically increases grouting pressure ;in, The gain coefficient is until... Meets the standards; Step S522, Geological and structural state perception; Internal sensing: In the rock mass surrounding and ahead of the tunnel, deploy or utilize advanced geological prediction systems to acquire in real time: microseismic / acoustic emission signals to monitor the development of rock mass fractures; deep displacement to monitor stratum slippage; and pore water pressure changes to monitor seepage field disturbances. Surface sensing: Inside the tunnel, convergence meters and total stations are used to automatically monitor tunnel convergence deformation; on the surface, surveying robots, GB-InSAR, or 3D laser scanners are used to continuously monitor changes in the surface settlement field.

8. The refined treatment decision-making method for complex geological disaster sources by fusing multi-source data according to claim 7, characterized in that, By comparing real-time sensed data with prediction models and control thresholds, dynamic optimization and adjustment of construction parameters can be achieved, specifically including: Step S531, Data Fusion and Status Diagnosis: The acquired multi-dimensional real-time data is integrated and analyzed in a unified data platform; the current construction status is diagnosed by comparing it in real time with the safety thresholds predicted by numerical simulation. Step S532: Based on the diagnostic results, the system generates control commands or issues warnings to operators and provides control suggestions: Shield tunneling parameter control: If the diagnosis finds that the rock mass ahead is weak and deformed, the system will automatically or prompt the operator to appropriately reduce the tunneling speed and increase the soil chamber pressure to stabilize the excavation face; Enhanced treatment parameters: If the diagnosis finds that the deformation in the grouting reinforcement area still exceeds expectations, the system instructs the intelligent grouting system to dynamically increase the grouting pressure or perform compensatory grouting in a specific area; Dynamic reinforcement of support: If the initial support stress or deformation is detected to reach the warning value, the system will prompt to strengthen the support parameters in the next cycle; Step S533, Risk Warning and Emergency Intervention: A multi-level early warning mechanism is established. When the risk level is determined to be orange or red after the fusion of multiple data, the system will not only adjust the parameters, but also automatically trigger an audible and visual alarm and set up an emergency shutdown interlock to force manual inspection and decision-making.

9. The refined treatment decision-making method for complex geological disaster sources by fusing multi-source data according to claim 1, characterized in that, In step S6, the effect evaluation and model update are carried out, and the specific process is as follows: Step S61: Objectively and quantitatively evaluate the effectiveness of the treatment plan by using the post-construction data collected. Step S611, posterior data collection; (1) Direct exposure verification includes: core drilling verification and in-hole television / optical imaging; (2) Non-destructive testing verification includes: cross-hole CT re-measurement and seismic wave / surface wave method; (3) Long-term stability monitoring: Continue to collect data on tunnel convergence, surface settlement and support structure stress for a period of time to confirm whether the deformation and stress have stabilized and are less than the control value; Step S612: Establish an effect evaluation index system based on posterior data, specifically including: (1) Filling percentage index: Based on core sampling and in-hole television, evaluate the percentage of voids / cracks filled by the grout; (2) Strength improvement rate: The ratio of rock mass strength / wave velocity value after treatment to the value before treatment; (3) Integrity index: the area reduction rate or wave velocity increase rate of the abnormal area in the CT image after treatment; (4) Stability indicators: deformation convergence rate and final stable value after treatment; Based on indicators (1)-(4), a quantitative conclusion of excellent, qualified, or unqualified is given for the overall treatment effect score; among which, the quantitative scoring formula for treatment effect is: ； in, Indicates the overall effect score; and These represent the elastic wave velocities before and after treatment, respectively. and These represent the areas of the low-velocity abnormality region in CT images before and after treatment, respectively. This represents the final displacement value after treatment; Indicates the allowable displacement value; , and These represent the weights for strength, integrity, and stability, respectively. Step S62: Dynamic updating of the three-dimensional geological model and numerical analysis model; Step S621: Update the geological attribute model; Disaster source model update: Update the attributes of the treated disaster source model from untreated to grout-filled or reinforced, and correct the geometric boundaries and physical and mechanical parameters based on the verification data; Rock mass parameter zoning update: Based on the strength improvement rate and wave velocity improvement rate, the rock mass quality grade and mechanical parameters of the reinforced area are zoning correction in the three-dimensional geological model; Step S622: Numerical analysis model correction; The updated 3D geological model was re-imported into the numerical simulation software. Using the actual deformation and stress data monitored after treatment as the target, the constitutive model parameters in the numerical model were inverted to make the numerical simulation results match the actual situation. Step S63: Self-learning and evolution of the knowledge base and case library; Successful treatment experiences are transformed into system-stored data as historical experience data. Step S631: Create a new successful treatment case, whose data structure includes: (1) Problem description: geological conditions, disaster source characteristics, and risk assessment level before treatment; (2) Treatment plan: All details of the optimal treatment plan to be adopted; (3) Results data: All posterior data collected and results evaluation conclusions; (4) Contextual information: Project name, geological type, hydrological conditions; Step S632: Knowledge rule optimization and generation; Validate and strengthen existing rules: If the current successful case highly matches a recommendation rule in the knowledge base, the confidence of that rule will be increased; Generate new rule: If a new optimized treatment plan is adopted and excellent results are achieved, the system will automatically or semi-automatically generate a new treatment rule. Failure Lessons Record: If the treatment is unsatisfactory, the case will be recorded in the database and the reasons will be analyzed, so as to issue a warning when similar situations are encountered in the future.

Images