A water-rich tunnel section construction quality safety risk management system and method

Abstract

This invention relates to the field of tunnel construction safety management technology, specifically a quality and safety risk management system and method for water-rich tunnel sections. The method includes: generating a three-dimensional geological cloud map from scanned data of the tunnel section to be constructed, and analyzing this data to form a set of blasting excavation parameters; collecting hydrogeological environmental data around the tunnel wall after blasting operations, establishing an initial seepage field equation, introducing a machine learning algorithm to nonlinearly fit groundwater pressure and rainfall parameters, and generating a dynamic seepage prediction model; comparing the predicted and measured seepage flow values, and determining support parameters based on the deviation and fracture zone distribution when the deviation exceeds a preset tolerance; collecting surrounding rock convergence displacement data and associating it with rock mass integrity indicators; and generating safety and quality control instructions including risk level classification. This method can dynamically adapt to hydrogeological changes, accurately match support parameters with surrounding rock risk characteristics, and achieve refined quality and safety management throughout the entire construction process.

Description

Technical Field

[0001] This invention relates to the field of tunnel construction safety management technology, and in particular to a quality and safety risk management system and method for water-rich tunnel sections. Background Technology

[0002] During tunnel and underground cavern construction, water-rich sections often face risks of sudden water inrush, mudslides, and surrounding rock instability due to fractured rock masses, well-developed joints and fissures, and significant impact from rainfall. Current technologies primarily rely on traditional methods such as geological sketching and advanced drilling for geological prediction, combined with empirical formulas to estimate water inflow, and determining support parameters based on the surrounding rock grade and relevant standards. These methods struggle to accurately depict the three-dimensional spatial distribution of complex geological structures, and seepage calculations often employ static models, failing to respond in real-time to changes in groundwater dynamics caused by rainfall infiltration, resulting in significant errors in water inflow prediction.

[0003] Existing support decision-making mechanisms typically solidify design parameters before construction, lacking adaptive adjustment methods on-site based on the differences between actual hydrogeological responses and predictive models. When the actual seepage environment deviates from the pre-design, it's impossible to quantify the degree of deviation in a timely manner and optimize the configuration of support structures such as anchor spraying, steel frames, and anchor bolts. This can easily lead to insufficient support causing collapses or excessive support resulting in resource waste. Furthermore, conventional monitoring data is often analyzed independently, without effectively establishing a correlation mapping with previous geological microstructure indicators, making it difficult to achieve dynamic and accurate risk level determination. A risk management method is needed that can integrate dynamic environmental factors and refined geological structural characteristics to achieve full-chain optimization from prediction to feedback control. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a construction quality and safety risk management system and method for water-rich tunnel sections.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a method for managing the construction quality and safety risks of water-rich tunnel sections, comprising:

[0006] The tunnel section to be constructed is scanned and data is collected to generate a three-dimensional geological cloud map;

[0007] The three-dimensional geological cloud map is analyzed to form a set of blasting excavation parameters;

[0008] After blasting operations are carried out based on the blasting excavation parameter set, hydrogeological environmental data around the tunnel wall are collected. The initial seepage field equation is established using numerical simulation methods, and machine learning algorithms are introduced to perform nonlinear fitting of groundwater pressure and rainfall parameters to generate a dynamic seepage volume prediction model that reflects the changes in seepage patterns.

[0009] The predicted seepage flow rate output by the dynamic seepage flow rate prediction model is compared with the actual seepage flow rate measured in real time. When the deviation between the predicted seepage flow rate and the actual seepage flow rate exceeds the preset tolerance, the thickness of the shotcrete, the spacing of the steel frame, and the arrangement density of the system anchor bolts are determined according to the magnitude of the deviation and the distribution of the fracture zone in the geological structure.

[0010] During the construction of the support structure, permanent monitoring points and deformation monitoring instruments are used to collect the surrounding rock convergence displacement data. The surrounding rock convergence displacement data is then correlated with the rock mass integrity index in the three-dimensional geological cloud map to generate safety and quality control instructions that include risk level classification.

[0011] As a further aspect of the present invention, the step of scanning and data acquisition of the tunnel section to be constructed to generate a three-dimensional geological cloud map includes:

[0012] Using the TSP advanced detection system, three-dimensional geological radar and intelligent borehole imaging detection technology, the topography, strata lithology and geological structure characteristics of the tunnel section to be constructed are scanned and data is collected to generate a three-dimensional geological cloud map containing rock mass integrity indicators and the distribution of potential fracture zones.

[0013] The three-dimensional geological cloud map is analyzed to form a set of blasting excavation parameters, including:

[0014] Combining the rock mass integrity index at the current working face location in the three-dimensional geological cloud map, the blasting vibration velocity data recorded by the blasting vibration monitoring instrument in the previous cycle is called up, and the dynamic response characteristics of the current rock mass are inverted through the surrounding rock stability analysis software. Based on this, the drilling angle and charge amount of the next cycle operation are dynamically adjusted to form a set of blasting excavation parameters after experimental adjustment.

[0015] The aforementioned technologies, including the TSP advanced detection system, 3D ground-penetrating radar, and intelligent borehole imaging detection, include:

[0016] The TSP advanced detection system emits seismic waves in front of the tunnel excavation face and receives reflected signals, using wave velocity differences to identify the location and scale of fault fracture zones.

[0017] Three-dimensional ground-penetrating radar was used to perform gridded scanning of the tunnel arch and sidewalls to obtain images of the dielectric constant distribution of shallow strata and identify loose bodies with abnormal water content.

[0018] The intelligent borehole imaging detection technology was used to conduct panoramic photography and acoustic testing on key boreholes to obtain the development of fractures in the borehole wall and core images.

[0019] By unifying the coordinates and fusing the seismic wave reflection data, dielectric constant distribution image, and borehole wall image, and removing noise data caused by instrument errors, a three-dimensional geological cloud map containing the physical and mechanical parameters of the rock mass is constructed.

[0020] As a further aspect of the present invention, combining the rock mass integrity index at the current working face location in the three-dimensional geological cloud map, the blasting vibration velocity data recorded by the blasting vibration monitoring instrument from the previous cycle is invoked, including:

[0021] Extract the rock mass integrity index within the radius of the center point coordinates of the current working face from the three-dimensional geological cloud map;

[0022] Retrieve the historical database stored in the blasting vibration monitoring instrument and search for the peak particle vibration velocity, dominant frequency, and vibration duration of the previous operating cycle that are similar to the current rock mass integrity index.

[0023] Calculate the correlation coefficient between particle vibration velocity and rock mass integrity index in the historical database. If the correlation coefficient is lower than the set threshold, introduce vibration data from adjacent working faces for interpolation correction.

[0024] As a further aspect of the present invention, the step of establishing the initial seepage field equation using numerical simulation methods and introducing machine learning algorithms to perform nonlinear fitting of groundwater pressure and rainfall parameters includes:

[0025] Based on the permeability coefficient and specific yield parameters in the hydrogeological survey report, the initial seepage field equations for the saturated and unsaturated regions around the tunnel are constructed using the finite difference method.

[0026] Groundwater pressure sensors and rain gauges are installed inside the tunnel to collect high-frequency time series data as input features, and the total seepage volume measured daily at regular intervals is used as the output label.

[0027] Long Short-Term Memory (LSTM) network was chosen as the main structure of the machine learning algorithm. The time series change rate of groundwater pressure and the cumulative effect of rainfall were used as two neuron nodes in the network input layer. The network weights were trained through backpropagation to generate a dynamic seepage prediction model.

[0028] As a further aspect of the present invention, determining the thickness of the shotcrete, the spacing of the steel frame, and the arrangement density of the system anchor bolts based on the magnitude of the deviation value and the distribution of fracture zones in the geological structure includes:

[0029] The positive and negative deviation ranges of the deviation value are set, each corresponding to a different level of water inrush risk;

[0030] When the deviation value is in the positive deviation range and the three-dimensional geological cloud map shows the existence of a continuous fracture zone, it is judged as a high risk of water inrush. At this time, the thickness of the anchor shotcrete is increased to the upper limit of the design, the spacing of the steel frame is reduced to the minimum allowable value, and the arrangement density of the system anchor bolts is increased within the influence range of the fracture zone.

[0031] When the deviation value is in the negative deviation range and the geological structure features show that it is an intact rock mass, it is judged as a low risk of water inrush. At this time, the support parameters of the standard design are adopted, and a ring-shaped drainage blind ditch and a longitudinal permeable blind pipe are added in the seepage section.

[0032] As a further aspect of the present invention, the targeted addition of annular drainage blind ditches and longitudinal permeable blind pipes in the seepage section includes:

[0033] An annular trench is excavated along the bottom outline of the tunnel. A geotextile filter layer is laid in the annular trench, and a permeable pipe is buried in the filter layer to form an annular drainage blind ditch. The outlet of the annular drainage blind ditch is directly connected to the longitudinal drainage gallery on the side of the tunnel.

[0034] In the longitudinal drainage gallery on both sides of the tunnel, longitudinal permeable blind pipes are laid parallel to the tunnel axis. The opening ratio of the pipe wall of the longitudinal permeable blind pipes is dynamically adjusted according to the seepage pressure of groundwater.

[0035] A water collection well is set up at the lowest point of the tunnel to introduce the water flow from the annular drainage blind ditch and the longitudinal permeable blind pipe into the water collection well. The accumulated water is then pumped to the sedimentation tank outside the tunnel by a drainage pumping station equipped with a submersible pump, thus realizing the closed-loop treatment of construction wastewater inside the tunnel.

[0036] As a further aspect of the present invention, the method of collecting surrounding rock convergence displacement data using permanent monitoring points and deformation monitoring instruments includes:

[0037] Multi-point displacement gauges and convergence gauges are installed at the tunnel arch, arch waist and sidewalls. The multi-point displacement gauges penetrate into the surrounding rock at different depths to monitor deep displacement of the surrounding rock, and the convergence gauges are used to monitor the relative displacement between two measuring points.

[0038] The data acquisition frequency is set to once per hour, and the surrounding rock convergence displacement data are continuously collected for no less than one complete hydrological cycle.

[0039] Temperature drift correction and zero-point drift correction are performed on the collected surrounding rock convergence displacement data to eliminate false displacements caused by instrument thermal expansion and contraction or installation stress release, and to generate a true surrounding rock deformation time series curve.

[0040] As a further aspect of the present invention, the surrounding rock convergence displacement data is correlated with the rock mass integrity index in the three-dimensional geological cloud map, including:

[0041] The actual surrounding rock deformation time series curve is sliced ​​according to time windows, and the displacement rate and cumulative displacement within each time window are extracted.

[0042] Extract rock mass integrity indicators and joint and fracture development directions at the corresponding monitoring points from the three-dimensional geological cloud map;

[0043] A mapping table between displacement rate and rock mass integrity index is established. When the displacement rate of any monitoring point increases continuously and the corresponding rock mass integrity index is lower than the critical value, the safety risk level of the area where the corresponding monitoring point is located is automatically raised by one level, and a corresponding safety quality control instruction is generated.

[0044] As a further aspect of the present invention, the generation of safety quality control instructions containing risk level classification includes:

[0045] Based on the consistency between the trend of the surrounding rock convergence displacement data and the geological structure characteristics, the construction safety risks are divided into four levels: stable, basically stable, unstable and unstable.

[0046] When the risk level is assessed as unstable, a quality control instruction is generated to suspend tunneling and immediately carry out radial grouting reinforcement.

[0047] When the risk level is assessed as basically stable, instructions are generated to reduce the tunneling speed and shorten the support closure distance;

[0048] When the risk level is assessed as stable, a production instruction is generated to maintain the existing construction parameters and continue operations.

[0049] As a further aspect of the present invention, the present invention also includes a construction quality and safety risk management system for water-rich tunnel sections. The system includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it implements the steps of the above-described method for construction quality and safety risk management of water-rich tunnel sections.

[0050] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0051] By constructing a dynamic seepage volume prediction model that integrates three-dimensional geological cloud maps, numerical simulation of seepage fields, and nonlinear fitting using machine learning, real-time coupled analysis of multiple environmental factors such as groundwater pressure and rainfall was achieved. This model overcomes the limitations of traditional static hydrogeological calculations, capturing the nonlinear evolution of the seepage field during rainfall infiltration, improving the timeliness and accuracy of water inflow prediction in water-rich tunnel sections, providing precise quantitative basis for drainage design and flood control measures during construction, and reducing construction interruptions and safety threats caused by sudden water and mud inflows.

[0052] Based on a threshold triggering mechanism for the deviation between predicted and measured seepage flow values, and combined with refined analytical results of the fracture zone distribution in the 3D geological cloud map, dynamic decisions are directly driven regarding the thickness of the shotcrete, the spacing of the steel frame, and the density of the system's anchor bolts. This parameter adjustment mode, which couples macroscopic seepage response differences with microscopic geological structural characteristics, eliminates the arbitrariness of human experience-based judgments, ensuring precise matching between the support structure configuration and the actual surrounding rock bearing capacity and the impact of the water environment. This guarantees cavern stability while avoiding material redundancy. Correlation analysis is performed between the surrounding rock convergence displacement data collected from permanent monitoring points and the rock mass integrity index in the 3D geological cloud map, establishing a mapping relationship from microscopic rock mass structural damage to macroscopic deformation response. This achieves adaptive risk level classification and closed-loop feedback for quality control. Attached Figure Description

[0053] Figure 1 The flowchart is a process for the construction quality and safety risk management method of the water-rich tunnel section described in this invention.

[0054] Figure 2 A flowchart for retrieving and correcting blasting vibration data;

[0055] Figure 3 A flowchart generated for a dynamic infiltration prediction model;

[0056] Figure 4 A biaxial analysis graph showing the increase in drainage efficiency and cost for different opening schemes;

[0057] Figure 5 The curve shows the comparison between the original data and the correction results of the displacement of the surrounding rock at a depth of 2 meters at the crown. Detailed Implementation

[0058] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0059] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0060] See Figure 1This method involves scanning and collecting data from the tunnel section to be constructed, generating a three-dimensional geological cloud map. Data analysis of the three-dimensional geological cloud map is used to form a set of blasting excavation parameters to guide on-site construction. After blasting operations based on this parameter set, hydrogeological environmental data around the tunnel wall are collected. Numerical simulation methods are used to establish an initial seepage field equation, and machine learning algorithms are introduced to perform nonlinear fitting of groundwater pressure and rainfall parameters, thereby generating a dynamic seepage prediction model that reflects changes in seepage patterns. The predicted seepage flow output by this model is compared with the measured seepage flow in real time. When the deviation exceeds a preset tolerance, the thickness of the shotcrete, the spacing of the steel frame, and the density of the system anchor bolts are determined based on the magnitude of the deviation and the distribution of fracture zones shown by the geological structural features in the three-dimensional geological cloud map. During the construction of the support structure, pre-deployed permanent monitoring points and deformation monitoring instruments are used to collect surrounding rock convergence displacement data. These displacement data are correlated with the rock mass integrity indicators in the three-dimensional geological cloud map, ultimately generating safety and quality control instructions that include risk level classification.

[0061] In one embodiment of the present invention, a TSP (Thunderbolt Spinning Point) advanced detection system, a three-dimensional ground-penetrating radar, and intelligent borehole imaging detection technology are used to scan and collect data on the topography, lithology, and geological structure of the tunnel section to be constructed. The TSP advanced detection system emits seismic waves in front of the tunnel face and receives reflected signals, using wave velocity differences to identify the location and scale of fault fracture zones. The three-dimensional ground-penetrating radar performs a gridded scan of the tunnel arch and sidewalls to obtain images of the dielectric constant distribution of shallow strata and identify loose bodies with abnormal water content. The intelligent borehole imaging detection technology is used to perform panoramic imaging and acoustic testing on key boreholes to obtain the fracture development status of the borehole wall and core images. The seismic wave reflection data, dielectric constant distribution images, and borehole wall images are coordinate unified and fused, and noise data caused by instrument errors is removed to construct a three-dimensional geological cloud map containing rock mass physical and mechanical parameters, rock mass integrity indicators, and the distribution of potential fracture zones. By combining the rock mass integrity index at the current working face location in the 3D geological cloud map, and retrieving the blasting vibration velocity data from the previous cycle recorded by the blasting vibration monitoring instrument, a blasting excavation parameter set is formed. The rock mass integrity index within the radius of the center point coordinates of the current working face is extracted from the 3D geological cloud map. The historical database stored in the blasting vibration monitoring instrument is retrieved to find the peak particle vibration velocity, dominant frequency, and vibration duration of the previous working cycle that are similar to the current rock mass integrity index. The correlation coefficient between the particle vibration velocity and the rock mass integrity index in the historical database is calculated. If the correlation coefficient is lower than a set threshold, vibration data from adjacent working faces are introduced for interpolation correction. The dynamic response characteristics of the current rock mass are inverted using surrounding rock stability analysis software, and the drilling angle and charge amount for the next cycle are dynamically adjusted accordingly, forming a test-adjusted blasting excavation parameter set.

[0062] In specific implementation, during the construction of a tunnel section of the Dianzhong Water Diversion Project, a comprehensive geological scan was conducted on the section of the tunnel to be constructed ahead of the excavation face. Specifically, the TSP (Through Seismic Detection System) was used to transmit seismic waves ahead of the tunnel face and receive reflected signals. The travel time difference and amplitude information between the received reflected waves and the direct waves were used to identify the location and scale of the fault fracture zone ahead by analyzing wave velocity differences. It can be understood that the reflection interface data obtained through the TSP provides information on the distribution of macroscopic geological structures. In some embodiments, a three-dimensional ground-penetrating radar (GPR) was used to perform a gridded scan of the tunnel arch and sidewalls, transmitting high-frequency electromagnetic waves and receiving reflected waves from the strata interface to obtain an image of the dielectric constant distribution of shallow strata. In specific implementation, areas in the dielectric constant distribution image obtained from the three-dimensional GPR scan with significantly higher dielectric constants than the surrounding rock mass were marked as loose bodies with abnormal water content. The intelligent borehole imaging detection technology was used to perform panoramic imaging and acoustic testing on key boreholes identified by the TSP advanced detection system and 3D ground-penetrating radar anomaly zones, obtaining high-resolution core images of the borehole wall fracture development. It can be understood that the images and acoustic data obtained through this intelligent borehole imaging detection technology provide intuitive and quantitative information about the rock mass structure surrounding the borehole.

[0063] In some embodiments, the seismic wave reflection data acquired by the TSP advanced detection system, the dielectric constant distribution image acquired by the 3D ground-penetrating radar, and the borehole wall image acquired by the borehole imaging intelligent detection technology are subjected to coordinate unification and data fusion. In specific implementation, spatial registration is performed on the multi-source data, mapping data points from different detection devices to a unified 3D tunnel coordinate system. A filtering algorithm is used to remove noise data caused by instrument errors, constructing a 3D geological cloud map containing rock mass physical and mechanical parameters, rock mass integrity indices, and the distribution of potential fracture zones. Combining the rock mass integrity indices at the current working face location in the 3D geological cloud map, the blasting vibration velocity data recorded by the blasting vibration monitoring instrument from the previous cycle is retrieved to form a blasting excavation parameter set. In specific implementation, the rock mass integrity indices within a 10-meter radius of the center point coordinates of the current working face are extracted from the 3D geological cloud map. The historical database stored in the blasting vibration monitoring instrument is retrieved to find the peak particle vibration velocity, dominant frequency, and vibration duration of the previous working cycle that are similar to the current rock mass integrity indices. Optionally, the correlation coefficient between peak particle vibration velocity and rock mass integrity index in the historical database is calculated. The formula for calculating the correlation coefficient is as follows:

[0064]

[0065] in: Represents the correlation coefficient. Indicates the first Peak particle vibration velocity recorded from historical data points. Indicates and The corresponding rock mass integrity index of the working face area. This represents the average value of the selected historical mass vibration velocity peak data. This represents the average value of the corresponding rock mass integrity index data. This indicates the number of historical data points selected. In practical implementation, if the calculated correlation coefficient... If the vibration data falls below a set threshold of 0.7, vibration monitoring data from adjacent working faces with similar geological conditions to the current working face are used for interpolation correction. Using surrounding rock stability analysis software, the corrected vibration data and current rock mass integrity indices are input to invert the dynamic response characteristics of the current rock mass, and the drilling angle and charge amount for the next cycle are dynamically adjusted accordingly. Optionally, the drilling angle is adjusted based on the inverted rock mass anisotropy coefficient, and the charge amount is adjusted based on the inverted dynamic compressive strength of the rock mass, ultimately forming a set of experimentally adjusted blasting excavation parameters.

[0066] In one embodiment of the present invention, an initial seepage field equation is established using numerical simulation, and a machine learning algorithm is introduced to nonlinearly fit groundwater pressure and rainfall parameters. The process is as follows: Based on the permeability coefficient and specific yield parameters in the hydrogeological survey report, the initial seepage field equations for the saturated and unsaturated areas around the tunnel are constructed using the finite difference method. Groundwater pressure sensors and rain gauges are deployed inside the tunnel to collect high-frequency time series data as input features, and the total seepage volume measured daily is used as the output label. A long short-term memory network is selected as the main structure of the machine learning algorithm, and the time series change rate of groundwater pressure and the cumulative effect of rainfall are used as two neuron nodes in the network input layer. The network weights are trained through the backpropagation algorithm to generate a dynamic seepage volume prediction model.

[0067] In practical implementation, during the construction of a water-rich tunnel in the Dianzhong Water Diversion Project, dynamic prediction of tunnel wall seepage is required after blasting excavation. Numerical simulation is used to establish the initial seepage field equations, and machine learning algorithms are introduced to perform nonlinear fitting of groundwater pressure and rainfall parameters. In practice, based on the permeability coefficient k and specific yield μ parameters of the soil and rock layers obtained from the preliminary hydrogeological survey report, the tunnel excavation profile is used as the boundary, and the surrounding area is spatially discretized using the finite difference method to construct the initial seepage field control equations, which include saturated and unsaturated zones. In some embodiments, the initial seepage field control equations can be expressed on a two-dimensional profile as follows:

[0068]

[0069] in: Used to represent the differentiation of a multivariable function with respect to a single variable. This represents the total head value at the calculation point. and Represents spatial coordinates, and These respectively represent the rock and soil mass in direction and The main permeability coefficient in the direction, Indicates water storage rate, The time dimension is used. It can be understood that the finite difference method discretizes the continuous seepage region into regular grid cells, and solves the above equations at each cell node to obtain the pressure head distribution of the initial seepage field. Groundwater pressure sensors and automatic rain gauges are deployed at key sections of the excavated section inside the tunnel to collect high-frequency time series data as input features for the machine learning model. In specific implementation, the groundwater pressure sensors record pore water pressure values ​​every ten minutes, and the automatic rain gauges record cumulative rainfall every hour. The total tunnel seepage volume measured through the catch weir at 8:00 AM daily is used as the output label for training the machine learning model. A long short-term memory network is selected as the main structure of the machine learning algorithm to construct a dynamic seepage volume prediction model.

[0070] In some embodiments, the input layer of the Long Short-Term Memory (LSTM) network has two neuron nodes: one node receives the time-series rate of change of groundwater pressure, and the other node receives the cumulative effect of rainfall. The time-series rate of change of groundwater pressure is calculated from the pore water pressure values ​​over multiple consecutive time steps, reflecting the dynamic trend of water pressure; the cumulative effect of rainfall is characterized by the accumulated rainfall within a set time window, reflecting the intensity of rainfall infiltration. Optionally, the LSM network has three hidden layers, each containing 50 neurons. Using a backpropagation algorithm, the connection weights and biases of the LSM network are iteratively trained with historically collected high-frequency time-series data and corresponding daily total infiltration labels to minimize the error between predicted and measured infiltration. In a specific implementation, mean squared error is used as the loss function during training, and the Adam optimizer is used to update the network parameters. Optionally, when the loss function no longer decreases significantly over multiple consecutive training cycles, the LSM network training is considered converged. At this point, the network weights are saved, generating a dynamic infiltration prediction model that can be used to predict infiltration in future periods.

[0071] In one embodiment of the present invention, the thickness of the shotcrete, the spacing of the steel frame, and the arrangement density of the system anchors are determined based on the magnitude of the deviation value and the distribution of fractured zones in the geological structure. Positive and negative deviation ranges are defined for the deviation value, corresponding to different levels of water inrush risk. When the deviation value is in the positive deviation range and a continuous fractured zone is shown in the three-dimensional geological cloud map, it is determined to be a high water inrush risk. In this case, the thickness of the shotcrete is increased to the design upper limit, the spacing of the steel frame is reduced to the minimum allowable value, and the arrangement density of the system anchors is increased within the influence range of the fractured zone. When the deviation value is in the negative deviation range and the geological structure shows intact rock mass, it is determined to be a low water inrush risk. In this case, standard design support parameters are used, and annular drainage blind ditches and longitudinal permeable blind pipes are specifically added in the seepage section.

[0072] In practical implementation, during the construction of a water-rich tunnel in the Dianzhong Water Diversion Project, the predicted seepage flow rate output by the dynamic seepage prediction model was continuously compared with the actual measured seepage flow rate monitored in real time. The support was adjusted based on the deviation and geological conditions. The thickness of the shotcrete, the spacing of the steel frame, and the density of the system's anchor bolts were determined according to the magnitude of the deviation and the distribution of fractured zones in the geological structure. In practical implementation, positive and negative deviation ranges were defined for the deviation values. In some embodiments, the deviation value was defined... The calculation formula is:

[0073]

[0074] in: This represents the predicted seepage flow rate output by the dynamic seepage prediction model. This represents the actual measured value of seepage volume monitored in real time within the same time period. This represents the relative deviation expressed as a percentage. It can be understood that a preset tolerance threshold is set. When the deviation value When the deviation value is within the positive deviation range, it is determined that the deviation value is within the positive deviation range; when the deviation value is within the positive deviation range, it is determined that the deviation value is within the positive deviation range. When the deviation value falls within the negative deviation range, the positive and negative deviation ranges correspond to different inrush risk levels. Optionally, a preset tolerance threshold can be used. The value is 10%.

[0075] In practical implementation, when the deviation value is in the positive deviation range and the 3D geological cloud map shows a continuous fracture zone, the system determines that the current tunnel section has a high risk of water inflow. At kilometer marker K50+120 of a certain section of the Dianzhong Water Diversion Project, the dynamic seepage prediction model output a predicted daily seepage flow of 150 cubic meters, while the actual measured seepage flow on site reached 180 cubic meters per day. The deviation value was calculated accordingly. The value is -16.7%, and its absolute value is greater than the preset tolerance threshold. ,and Therefore, it falls within the negative deviation range. However, regarding the positive deviation range, if the predicted value is 180 cubic meters per day while the actual measured value is 150 cubic meters per day, then... greater than Furthermore, the 3D geological cloud map shows a fault-connected fractured zone with a width exceeding 3 meters at this mileage location. In this case, the thickness of the shotcrete is increased to the design limit of 30 cm, the spacing of the steel arches is reduced to the minimum allowable value of 0.5 meters, and the density of the system anchors is increased to 1.5 anchors per square meter within a 10-meter radius before and after the fractured zone's influence. In some embodiments, after determining a high risk of water inflow, the adjustment of the support parameters is based on the magnitude of the deviation value and the scale and occurrence of the fractured zone, and specific values ​​are directly provided by querying the pre-stored design parameter database. When the deviation value is in the negative deviation range and the geological structure shows intact rock mass, the system determines the current tunnel section to have a low risk of water inflow. At mileage K50+300 of a certain section of the Dianzhong Water Diversion Project, the deviation value... The calculated value is -5%, falling within the negative deviation range. Furthermore, the geological structural features extracted from the 3D geological cloud map show that the rock mass in this area is intact, with a high rock mass integrity index and no obvious structural fracturing. Therefore, it is understandable that standard design support parameters are used at this point, namely, 20 cm thick shotcrete, steel arch frames spaced 1.2 meters apart, and a system of 0.8 anchor bolts per square meter. Optionally, a ring-shaped drainage ditch and longitudinal permeable blind pipes can be specifically added to the seepage section. In some embodiments, the purpose of adding drainage measures is to divert rather than block, in order to reduce the water pressure behind the lining.

[0076] In one embodiment of the invention, a ring-shaped drainage ditch and longitudinal permeable blind pipes are specifically added to the seepage section. This includes: excavating a ring-shaped trench along the tunnel bottom outline; laying a geotextile filter layer within the ring-shaped trench; and burying permeable pipes within the filter layer to form a ring-shaped drainage ditch. The outlet of the ring-shaped drainage ditch is directly connected to the longitudinal drainage gallery on the side of the tunnel. Longitudinal permeable blind pipes are laid parallel to the tunnel axis within the longitudinal drainage gallery on both sides of the tunnel. The porosity of the pipe walls of the longitudinal permeable blind pipes is dynamically adjusted according to the groundwater seepage pressure. A collection well is installed at the lowest point of the tunnel to guide the water flow collected from the ring-shaped drainage ditch and the longitudinal permeable blind pipes into the collection well. The accumulated water is then pumped to a sedimentation tank outside the tunnel by a drainage pumping station equipped with a submersible pump.

[0077] In the specific implementation, in the low-risk water inrush section from K50+300 to K50+400 of a water-rich tunnel in the Dianzhong Water Diversion Project, a ring-shaped drainage ditch and longitudinal permeable pipes were specifically added. The addition of these ditches in the seepage section involved excavating a ring-shaped trench along the tunnel's bottom outline. In practice, the trench was 0.5 meters deep and 0.4 meters wide, with its direction strictly parallel to the tunnel's bottom design outline. A 400g / m² needle-punched non-woven geotextile was laid as a filter layer within the excavated ring-shaped trench. A 150mm diameter flexible permeable pipe was then buried on top of the geotextile filter layer to form the ring-shaped drainage ditch. The outlet of the ring-shaped drainage ditch was directly connected to the completed longitudinal drainage gallery on the side of the tunnel via a pre-embedded PVC connecting pipe. It is understandable that the geotextile filter layer wrapping the flexible permeable pipe can effectively prevent silt particles from groundwater from entering the permeable pipe and causing blockage. The slope of the annular drainage blind ditch is consistent with the longitudinal slope of the tunnel floor to ensure smooth drainage. Longitudinal permeable blind pipes are laid parallel to the tunnel axis within the longitudinal drainage galleries on both sides of the tunnel. In specific implementation, the longitudinal permeable blind pipes use HDPE perforated corrugated pipes with an outer diameter of 100 mm and a wall thickness of 5 mm. The porosity of the longitudinal permeable blind pipe wall is dynamically adjusted according to the groundwater seepage pressure. The relationship between the porosity and seepage pressure is determined by the following formula:

[0078]

[0079] in: This indicates the porosity of the longitudinal permeable blind pipe wall. This represents the seepage water pressure at the tunnel sidewall, as measured by sensors. The reference pressure constant, with a value of 1 kPa, is used to make the parameters within the logarithmic function... Become a dimensionless quantity. and This is a dimensionless adjustment factor related to the permeability of the surrounding rock and drainage requirements. In some embodiments, for the water-rich section of the red mudstone interbedded with sandstone in central Yunnan, the coefficient is... The value is 2.5, and the coefficient is... The value is 1.5. This can be understood as osmotic pressure. The larger the value, the higher the calculated aperture ratio. The size should be increased accordingly to ensure sufficient drainage capacity. See Table 1 for the opening schemes used in different permeable pressure zones for longitudinal permeable blind pipes.

[0080] Table 1: Configuration Table of Longitudinal Permeable Blind Pipe Opening Ratio Osmotic pressure P (kilopascals) Calculate the opening ratio η (%) The actual solution adopted is the hole-opening scheme. P<50 η<11.5 Standard opening, 8mm diameter, 200mm hole spacing 50≤P<100 11.5≤η<14.9 Encrypted Level 1 opening, 8mm diameter, 150mm hole spacing 100≤P<150 14.9≤η<17.4 Encrypted secondary aperture, 8mm diameter, 100mm hole spacing. P≥150 η≥17.4 Three-level encryption aperture, 10mm diameter, 100mm hole spacing

[0081] In practical implementation, longitudinal permeable blind pipes are laid at the bottom of the longitudinal drainage corridor and connected to the outlet of the annular drainage blind ditch via tee fittings, enabling the two drainage systems to be interconnected. A sump is installed at the lowest point of the tunnel's longitudinal slope to guide the water flow collected from the annular drainage blind ditch and the longitudinal permeable blind pipes into the sump. In some embodiments, the sump is 3 meters long, 2 meters wide, and 2 meters deep, with its inner wall lined with concrete to prevent seepage. Optionally, a drainage pumping station equipped with a submersible pump pumps the accumulated water in the sump to an external sedimentation tank. It is understood that the external sedimentation tank allows for natural sedimentation and preliminary purification of the pumped-out construction wastewater, achieving closed-loop treatment of the construction wastewater inside the tunnel.

[0082] See Figure 4 In the analysis of drainage efficiency and cost of different opening schemes for longitudinal permeable blind pipes in the water-rich tunnel section, a dual-axis visualization system was adopted to achieve a synergistic evaluation of performance and economy. The left vertical axis represents drainage efficiency (%), and a bar chart is used to show the improvement in drainage capacity under different opening schemes: the standard opening scheme is used as the benchmark scheme, with a drainage efficiency of 75%; the first-level densified scheme increases to 85%, the second-level densified scheme further increases to 92%, and the third-level densified scheme reaches a peak of 95%, showing a monotonically increasing trend with the increase of opening density and pore size, which is consistent with the design logic of a positive correlation between seepage pressure and drainage capacity. The right vertical axis represents the percentage increase in cost (%), and a line chart is used to show the cost increase of each scheme relative to the benchmark scheme: the cost increase of the standard opening scheme is 0%, and the cost increases of the first-level, second-level, and third-level densified schemes are 10%, 20%, and 35%, respectively. The cost growth rate is significantly faster than the drainage efficiency growth rate, reflecting the engineering characteristics of diminishing marginal benefits—although the third-level densified scheme achieves the highest drainage efficiency, the cost increase has exceeded the drainage efficiency increase, and the economy has significantly decreased. From a techno-economic perspective, the secondary-level densification scheme achieves a better balance between drainage efficiency (92%) and cost increase (20%): compared with the primary-level densification scheme, the drainage efficiency is improved by 7 percentage points while the cost increases by only 10 percentage points; compared with the tertiary-level densification scheme, only 3 percentage points of drainage efficiency are sacrificed to reduce the cost increase by 15 percentage points, making it more suitable for engineering needs in the water-rich section of the Dianzhong Red Bed with a seepage pressure range of 100-150 kPa.

[0083] In one embodiment of the present invention, during the construction of the support structure, permanent monitoring points and deformation monitoring instruments are used to collect surrounding rock convergence displacement data. Multi-point displacement gauges and convergence gauges are installed at the tunnel arch, arch waist, and sidewalls. The multi-point displacement gauges penetrate to different depths within the surrounding rock to monitor deep displacement, while the convergence gauges monitor the relative displacement between two measuring points. The data acquisition frequency is set to once per hour, continuously collecting surrounding rock convergence displacement data for at least one complete hydrological cycle. Temperature drift correction and zero-point drift correction are performed on the collected surrounding rock convergence displacement data to eliminate false displacement values ​​caused by instrument thermal expansion and contraction or installation stress release, generating a true surrounding rock deformation time-series curve. The surrounding rock convergence displacement data is correlated with rock mass integrity indicators in a three-dimensional geological cloud map. The true surrounding rock deformation time-series curve is sliced ​​according to time windows, and the displacement rate and cumulative displacement within each time window are extracted. Rock mass integrity indicators and joint and fracture development directions corresponding to the monitoring point locations are extracted from the three-dimensional geological cloud map. A mapping table between displacement rate and rock mass integrity index is established. When the displacement rate at any monitoring point continuously increases and the corresponding rock mass integrity index falls below the critical value, the safety risk level of the area where the corresponding monitoring point is located is automatically raised by one level. Safety and quality control instructions containing risk level classifications are generated. Based on the consistency between the changing trend of the surrounding rock convergence displacement data and the geological structural characteristics, construction safety risks are classified into four levels: stable, basically stable, unstable, and instable. When the risk level is assessed as unstable, a quality control instruction to suspend tunneling and immediately carry out radial grouting reinforcement is generated. When the risk level is assessed as basically stable, an instruction to reduce the tunneling speed and shorten the support closure distance is generated. When the risk level is assessed as stable, a production instruction to maintain the existing construction parameters and continue operation is generated.

[0084] In the specific implementation, during the construction of the support structure of the tunnel section near kilometer marker K60+100 in a certain section of the Dianzhong Water Diversion Project, permanent monitoring points and deformation monitoring instruments were used to collect surrounding rock convergence displacement data. A monitoring network was formed by installing multi-point displacement gauges and convergence gauges at the tunnel arch, arch waist, and sidewalls. The probes of the multi-point displacement gauges penetrated into the surrounding rock at three different depths of two, four, and six meters to monitor displacement changes at different levels within the deeper layers of the surrounding rock. The convergence gauges were fixed between measuring piles installed on the tunnel wall to monitor the relative displacement between two measuring points. The data acquisition frequency was set to once per hour, and the automatic recorder continuously collected surrounding rock convergence displacement data for no less than one complete hydrological cycle (thirty days). Temperature drift correction and zero-point drift correction are performed on the collected raw surrounding rock convergence displacement data. In practice, temperature drift correction uses the temperature sensor reading built into the instrument and corrects the raw reading using the temperature-displacement compensation coefficient calibrated by the instrument at the factory. Zero-point drift correction reads the baseline reading of the instrument under stable no-load conditions at the beginning of each data acquisition cycle and subtracts it from subsequent data as the zero-point offset, thereby eliminating false displacement caused by thermal expansion and contraction of the instrument or stress release during installation, and generating a surrounding rock deformation time series curve that can truly reflect the mechanical behavior of the surrounding rock. The surrounding rock convergence displacement data is correlated with the rock mass integrity index in the 3D geological cloud map. In practice, the real surrounding rock deformation time series curve is sliced ​​into 24-hour time windows, and the displacement rate and cumulative displacement within each time window are extracted. The rock mass integrity index and the dominant development direction of joints and fractures at the corresponding monitoring point coordinates are extracted from the 3D geological cloud map. In some embodiments, a quantitative mapping relationship between displacement rate and rock mass integrity index is established, and its expression is:

[0085]

[0086] in, This represents the displacement rate calculated from the surrounding rock deformation time series curve. Indicates the reference displacement rate. This represents the rock mass integrity index extracted from the 3D geological cloud map. This represents the attenuation coefficient describing the degree to which rock mass integrity inhibits deformation rate. It is a natural constant. This can be understood as the right side of the formula... The overall dimension is the reference displacement rate. Dimensions , with the displacement rate on the left The dimensions are consistent. When the displacement rate at any monitoring point The rock mass integrity index obtained from the three-dimensional geological cloud map continued to increase over three consecutive time windows. When the risk level falls below the critical value of 0.55, the monitoring system automatically raises the safety risk level of the area where the corresponding monitoring point is located by one level and generates a corresponding safety quality control instruction. This instruction, which includes risk level classifications, is implemented by classifying construction safety risks into four levels: stable, basically stable, unstable, and instable, based on the correlation between the trend of the surrounding rock convergence displacement data and the geological structural features revealed by the 3D geological cloud map. When the risk level is assessed as unstable by the system, a quality control instruction is generated to suspend excavation and immediately perform radial grouting reinforcement on the convergence deformation area. In some embodiments, the radial grouting reinforcement instruction includes specific parameters such as the circumferential spacing of the grouting holes, the hole depth, and the grouting pressure. When the risk level is assessed as basically stable, an instruction is generated to reduce the excavation speed to 50% of the original plan and shorten the support closure distance to within 25 meters of the working face. The support closure distance refers to the length from the excavation face to the initial support ring formation position. When the risk level is assessed as stable, a production instruction is generated to maintain the existing excavation progress, blasting parameters, and support construction parameters for continued operation. Optionally, all generated safety and quality control instructions are automatically pushed to the mobile terminals of on-site technicians and work teams via the construction management platform.

[0087] See Figure 5 In the monitoring and correction analysis of surrounding rock displacement in the Fushuidong section, the time-series comparison of the original displacement data and the corrected displacement data intuitively reflects the effect of eliminating instrument drift errors. The solid line in the figure represents the original displacement data containing drift errors. Its curve is generally high and the fluctuations are superimposed with false displacement components introduced by factors such as thermal expansion and contraction of the instrument and stress release during installation. The dashed line represents the true surrounding rock displacement data after temperature drift correction and zero-point drift correction. The influence of thermal expansion and contraction is corrected by the reading of the built-in temperature sensor and the temperature-displacement compensation coefficient calibrated by the factory. The zero-point offset is deducted from the reference reading under stable no-load conditions. Finally, the instrument system error is eliminated, and the true time-series characteristics of the mechanical deformation of the surrounding rock are restored. From the curve evolution trend, both types of data show a gradual increase with the increase of monitoring time, which is consistent with the deformation development characteristics of stress redistribution in the surrounding rock after tunnel excavation. However, the baseline of the corrected data is closer to the true initial state (0 displacement baseline), and the high-frequency fluctuations more purely reflect the deformation response of the surrounding rock itself, providing a reliable data basis for subsequent displacement rate calculation, rock mass integrity index correlation analysis and safety risk level classification.

[0088] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for managing the construction quality and safety risks of water-rich tunnel sections, characterized in that, include: The tunnel section to be constructed is scanned and data is collected to generate a three-dimensional geological cloud map; The three-dimensional geological cloud map is analyzed to form a set of blasting excavation parameters; After blasting operations are carried out based on the blasting excavation parameter set, hydrogeological environmental data around the tunnel wall are collected. The initial seepage field equation is established using numerical simulation methods, and machine learning algorithms are introduced to perform nonlinear fitting of groundwater pressure and rainfall parameters to generate a dynamic seepage volume prediction model that reflects the changes in seepage patterns. The predicted seepage flow rate output by the dynamic seepage flow rate prediction model is compared with the actual seepage flow rate measured in real time. When the deviation between the predicted seepage flow rate and the actual seepage flow rate exceeds the preset tolerance, the thickness of the shotcrete, the spacing of the steel frame, and the arrangement density of the system anchor bolts are determined according to the magnitude of the deviation and the distribution of the fracture zone in the geological structure. During the construction of the support structure, permanent monitoring points and deformation monitoring instruments are used to collect the surrounding rock convergence displacement data. The surrounding rock convergence displacement data is then correlated with the rock mass integrity index in the three-dimensional geological cloud map to generate safety and quality control instructions that include risk level classification.

2. The method for managing the construction quality and safety risks of a water-rich tunnel section according to claim 1, characterized in that, The process of scanning and data acquisition of the tunnel section to be constructed, generating a three-dimensional geological cloud map, includes: Using the TSP advanced detection system, three-dimensional geological radar and intelligent borehole imaging detection technology, the topography, strata lithology and geological structure characteristics of the tunnel section to be constructed are scanned and data is collected to generate a three-dimensional geological cloud map containing rock mass integrity indicators and the distribution of potential fracture zones. The three-dimensional geological cloud map is analyzed to form a set of blasting excavation parameters, including: Combining the rock mass integrity index at the current working face location in the three-dimensional geological cloud map, the blasting vibration velocity data recorded by the blasting vibration monitoring instrument in the previous cycle is called up, and the dynamic response characteristics of the current rock mass are inverted through the surrounding rock stability analysis software. Based on this, the drilling angle and charge amount of the next cycle operation are dynamically adjusted to form a set of blasting excavation parameters after experimental adjustment. The aforementioned technologies, including the TSP advanced detection system, 3D ground-penetrating radar, and intelligent borehole imaging detection, include: The TSP advanced detection system emits seismic waves in front of the tunnel excavation face and receives reflected signals, using wave velocity differences to identify the location and scale of fault fracture zones. Three-dimensional ground-penetrating radar was used to perform gridded scanning of the tunnel arch and sidewalls to obtain images of the dielectric constant distribution of shallow strata and identify loose bodies with abnormal water content. The intelligent borehole imaging detection technology was used to conduct panoramic photography and acoustic testing on key boreholes to obtain the development of fractures in the borehole wall and core images. By unifying the coordinates and fusing the seismic wave reflection data, dielectric constant distribution image, and borehole wall image, and removing noise data caused by instrument errors, a three-dimensional geological cloud map containing the physical and mechanical parameters of the rock mass is constructed.

3. The method for managing the construction quality and safety risks of a water-rich tunnel section according to claim 2, characterized in that, Based on the rock mass integrity index at the current working face location in the aforementioned 3D geological cloud map, the blasting vibration velocity data from the previous cycle recorded by the blasting vibration monitoring instrument is retrieved, including: Extract the rock mass integrity index within the radius of the center point coordinates of the current working face from the three-dimensional geological cloud map; Retrieve the historical database stored in the blasting vibration monitoring instrument and search for the peak particle vibration velocity, dominant frequency, and vibration duration of the previous operating cycle that are similar to the current rock mass integrity index. Calculate the correlation coefficient between particle vibration velocity and rock mass integrity index in the historical database. If the correlation coefficient is lower than the set threshold, introduce vibration data from adjacent working faces for interpolation correction.

4. The method for managing the construction quality and safety risks of a water-rich tunnel section according to claim 1, characterized in that, The method of establishing the initial seepage field equation using numerical simulation and introducing machine learning algorithms to perform nonlinear fitting of groundwater pressure and rainfall parameters includes: Based on the permeability coefficient and specific yield parameters in the hydrogeological survey report, the initial seepage field equations for the saturated and unsaturated regions around the tunnel are constructed using the finite difference method. Groundwater pressure sensors and rain gauges are installed inside the tunnel to collect high-frequency time series data as input features, and the total seepage volume measured daily at regular intervals is used as the output label. Long Short-Term Memory (LSTM) network was chosen as the main structure of the machine learning algorithm. The time series change rate of groundwater pressure and the cumulative effect of rainfall were used as two neuron nodes in the network input layer. The network weights were trained through backpropagation to generate a dynamic seepage prediction model.

5. The method for managing the construction quality and safety risks of a water-rich tunnel section according to claim 1, characterized in that, The determination of the thickness of the shotcrete, the spacing of the steel frame, and the arrangement density of the system anchor bolts based on the magnitude of the deviation value and the distribution of fracture zones in the geological structure includes: The positive and negative deviation ranges of the deviation value are set, each corresponding to a different level of water inrush risk; When the deviation value is in the positive deviation range and the three-dimensional geological cloud map shows the existence of a continuous fracture zone, it is judged as a high risk of water inrush. At this time, the thickness of the anchor shotcrete is increased to the upper limit of the design, the spacing of the steel frame is reduced to the minimum allowable value, and the arrangement density of the system anchor bolts is increased within the influence range of the fracture zone. When the deviation value is in the negative deviation range and the geological structure features show that it is an intact rock mass, it is judged as a low risk of water inrush. At this time, the support parameters of the standard design are adopted, and a ring-shaped drainage blind ditch and a longitudinal permeable blind pipe are added in the seepage section.

6. The method for managing the construction quality and safety risks of a water-rich tunnel section according to claim 5, characterized in that, The specific addition of annular drainage blind ditches and longitudinal permeable blind pipes in the seepage section includes: An annular trench is excavated along the bottom outline of the tunnel. A geotextile filter layer is laid in the annular trench, and a permeable pipe is buried in the filter layer to form an annular drainage blind ditch. The outlet of the annular drainage blind ditch is directly connected to the longitudinal drainage gallery on the side of the tunnel. In the longitudinal drainage gallery on both sides of the tunnel, longitudinal permeable blind pipes are laid parallel to the tunnel axis. The opening ratio of the pipe wall of the longitudinal permeable blind pipes is dynamically adjusted according to the seepage pressure of groundwater. A water collection well is set up at the lowest point of the tunnel to introduce the water flow from the annular drainage blind ditch and the longitudinal permeable blind pipe into the water collection well. The accumulated water is then pumped to the sedimentation tank outside the tunnel by a drainage pumping station equipped with a submersible pump, thus realizing the closed-loop treatment of construction wastewater inside the tunnel.

7. The method for managing the construction quality and safety risks of a water-rich tunnel section according to claim 1, characterized in that, The method of collecting surrounding rock convergence displacement data using permanent monitoring points and deformation monitoring instruments includes: Multi-point displacement gauges and convergence gauges are installed at the tunnel arch, arch waist and sidewalls. The multi-point displacement gauges penetrate into the surrounding rock at different depths to monitor deep displacement of the surrounding rock, and the convergence gauges are used to monitor the relative displacement between two measuring points. The data acquisition frequency is set to once per hour, and the surrounding rock convergence displacement data are continuously collected for no less than one complete hydrological cycle. Temperature drift correction and zero-point drift correction are performed on the collected surrounding rock convergence displacement data to eliminate false displacements caused by instrument thermal expansion and contraction or installation stress release, and to generate a true surrounding rock deformation time series curve.

8. The method for managing the construction quality and safety risks of a water-rich tunnel section according to claim 1, characterized in that, The surrounding rock convergence displacement data is correlated with the rock mass integrity index in the three-dimensional geological cloud map, including: The actual surrounding rock deformation time series curve is sliced ​​according to time windows, and the displacement rate and cumulative displacement within each time window are extracted. Extract rock mass integrity indicators and joint and fracture development directions at the corresponding monitoring points from the three-dimensional geological cloud map; A mapping table between displacement rate and rock mass integrity index is established. When the displacement rate of any monitoring point increases continuously and the corresponding rock mass integrity index is lower than the critical value, the safety risk level of the area where the corresponding monitoring point is located is automatically raised by one level, and a corresponding safety quality control instruction is generated.

9. The method for managing the construction quality and safety risks of a water-rich tunnel section according to claim 1, characterized in that, The generation of safety and quality control instructions containing risk level classifications includes: Based on the consistency between the trend of the surrounding rock convergence displacement data and the geological structure characteristics, the construction safety risks are divided into four levels: stable, basically stable, unstable and unstable. When the risk level is assessed as unstable, a quality control instruction is generated to suspend tunneling and immediately carry out radial grouting reinforcement. When the risk level is assessed as basically stable, instructions are generated to reduce the tunneling speed and shorten the support closure distance; When the risk level is assessed as stable, a production instruction is generated to maintain the existing construction parameters and continue operations.

10. A construction quality and safety risk management system for water-rich tunnel sections, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for managing the construction quality and safety risks of a water-rich tunnel section as described in any one of claims 1 to 9.

Images