Karst fissure water treatment method, device, terminal equipment and storage medium

Abstract

The application discloses a karst fissure water treatment method, device, terminal equipment and storage medium, and belongs to the karst fissure water treatment field, and the method is: obtaining a pressure field based on a to-be-interpolated point set, a water pressure measuring point and water pressure data, and obtaining a flow velocity field based on the pressure field and preset water parameters; obtaining a structural strain field based on the to-be-interpolated point set, a strain measuring point and strain data; obtaining a karst fissure water characteristic vector based on target monitoring data, the pressure field, the flow velocity field and the structural strain field; inputting the target monitoring data, the karst fissure water characteristic vector, the pressure field, the flow velocity field and the structural strain field into a karst fissure water risk model to obtain a karst fissure water risk grade and a contribution degree sequence; obtaining a target treatment scheme based on the target monitoring data, the karst fissure water risk grade, the contribution degree sequence, the pressure field, the flow velocity field and the structural strain field; and treating a construction influence area based on the target treatment scheme, so that the reliability of karst fissure water treatment is improved.

Description

Technical Field

[0001] This invention relates to the field of karst fissure water treatment, and more particularly to a method, apparatus, terminal equipment, and storage medium for karst fissure water treatment. Background Technology

[0002] In karst formations, karst fissure water is characterized by its high degree of concealment, rapid dynamic changes, and sudden onset. Conventional methods for treating karst fissure water mainly fall into two categories: one is large-area dewatering wells to lower the groundwater level and control seepage; the other is unified grouting and sealing, which involves grouting the suspected seepage areas as a whole or semi-whole.

[0003] However, with the increasing development of underground transportation hubs, the cut-and-cover method is widely used in urban subway stations and underground integrated transportation hubs to effectively reduce the impact on surface traffic and the environment. However, in the cut-and-cover environment, the aforementioned methods for treating karst fissure water have significant drawbacks: during subway cut-and-cover construction, karst fissure water is highly concealed, prone to sudden occurrences, and changes rapidly. Furthermore, the construction space is limited, and traditional dewatering and uniform grouting methods are poorly adaptable, easily leading to problems such as water inrush, sand inrush, and structural safety risks. Summary of the Invention

[0004] This invention provides a method, apparatus, terminal equipment, and storage medium for treating karst fissure water, which can solve the above-mentioned problems in the prior art and improve the reliability of karst fissure water treatment.

[0005] This invention provides a method for treating karst fissure water, comprising: Acquire target monitoring data and a set of interpolation points in the construction-affected area; Water pressure data and strain data are obtained based on the target monitoring data; The pressure field is obtained based on the set of interpolation points, the preset water pressure measurement points and the water pressure data, and the flow velocity field is obtained based on the pressure field and the preset water parameters. The structural strain field is obtained based on the set of interpolation points, the preset strain measurement points, and the strain data. Based on the target monitoring data, pressure field, flow velocity field, and structural strain field, the characteristic vector of karst fissure water is obtained; The target monitoring data, karst fissure water feature vector, pressure field, flow velocity field and structural strain field are input into the pre-acquired karst fissure water risk model to obtain the karst fissure water risk level and the ranking of karst fissure water feature contribution. Based on the target monitoring data, the risk level of karst fissure water, the ranking of the contribution of karst fissure water characteristics, the pressure field, the flow velocity field, and the structural strain field, a target treatment plan is obtained; The construction-affected area is treated based on the target treatment plan.

[0006] In the above scheme, by acquiring target monitoring data and a set of interpolation points in the construction-affected area, the pressure field, flow velocity field, and structural strain field are reconstructed, and the discrete measurement point information is reconstructed into a continuous spatial field. This overcomes the defect that limited measurement points cannot reflect the overall spatial distribution, providing a spatial quantitative basis for determining the subsequent target treatment scheme, thus obtaining a more accurate and reliable target treatment scheme. Then, based on this, the karst fissure water feature vector is extracted, and the target monitoring data, feature vector, and reconstructed field are jointly input into the pre-acquired karst fissure water risk model to obtain the accurate karst fissure water risk level and the ranking of karst fissure water feature contribution. This realizes intelligent classification and risk tracing under multi-source information fusion. Then, based on the target monitoring data, the karst fissure water risk level, the ranking of karst fissure water feature contribution, the pressure field, flow velocity field, and structural strain field, the target treatment scheme is obtained. Based on the target treatment scheme, the construction-affected area is treated, thereby effectively improving the reliability of karst fissure water treatment.

[0007] Furthermore, in the process of acquiring the target monitoring data and the set of interpolation points for the construction impact area, the step of acquiring the target monitoring data includes: Obtain the original monitoring data of the construction-affected area; The original monitoring data is subjected to outlier detection based on a preset outlier detection algorithm to obtain the first intermediate monitoring data. The first intermediate monitoring data is imputed for missing values ​​based on a preset missing value imputation algorithm to obtain the second intermediate monitoring data. The target monitoring data is obtained based on the preset dynamic time warping algorithm and the second intermediate monitoring data.

[0008] In the above scheme, high-quality target monitoring data is obtained by performing outlier detection, missing value interpolation and dynamic time warping alignment on the original monitoring data. This ensures the accuracy of subsequent pressure field and strain field reconstruction and feature extraction, reduces the interference of sensor noise and missing data on risk assessment, and improves the robustness of the system in complex construction environments.

[0009] Furthermore, obtaining the karst fissure water feature vector based on the target monitoring data, pressure field, flow velocity field, and structural strain field includes: Based on the target monitoring data, a subset of the measurement point signal features is obtained; A subset of spatial gradient features is obtained based on the pressure field and flow velocity field; Based on the structural strain field and the target monitoring data, a multi-source coupling feature subset is obtained; The feature vector of karst fissure water is obtained by splicing together the feature subsets of the measuring point signals, the spatial gradient feature subsets, and the multi-source coupling feature subsets.

[0010] In the above scheme, by extracting the feature subsets of the measuring point signals, the feature subsets of the spatial gradient, and the feature subsets of the multi-source coupling and splicing them into the feature vector of karst fissure water, a comprehensive condensation of the original monitoring data, the reconstructed field, and the structure-seepage coupling information is achieved. This enables the risk model to simultaneously perceive local temporal dynamics, spatial seepage gradient, and seepage-structure interaction, thereby enhancing the multi-dimensional information fusion capability of risk assessment.

[0011] Furthermore, the karst fissure water risk model includes a spatial attention module, a temporal attention module, and a multi-scale feature fusion layer. The step of inputting the target monitoring data, karst fissure water feature vectors, pressure field, flow velocity field, and structural strain field into the pre-acquired karst fissure water risk model to obtain the karst fissure water risk level and the ranking of karst fissure water feature contributions includes: The target monitoring data, karst fissure water feature vector, pressure field, flow velocity field, and structural strain field are input into the karst fissure water risk model. This allows the spatial attention module to obtain spatial attention weighted field features and spatial contribution ranking based on a preset graph attention network, the pressure field, flow velocity field, and structural strain field. The temporal attention module then obtains temporal attention weighted sequence features and temporal contribution ranking based on the target monitoring data. Finally, the multi-scale feature fusion layer nonlinearly fuses the karst fissure water feature vector, the spatial attention weighted field features, and the temporal attention weighted sequence features to obtain a comprehensive risk index. The risk level of the karst fissure water is obtained based on the comprehensive risk index, and the characteristic contribution ranking of the karst fissure water is obtained based on the spatial contribution ranking and the temporal contribution ranking.

[0012] In the above scheme, the spatial attention module is used to output spatial attention weighted field features and spatial contribution ranking based on pressure field, flow velocity field and structural strain field. The temporal attention module is used to output temporal attention weighted sequence features and temporal contribution ranking based on target monitoring data. Then, the comprehensive risk index is obtained through nonlinear fusion of multi-scale feature fusion layer, and the contribution ranking of karst fissure water features is derived. This realizes the spatiotemporal positioning and feature-level source tracing of risk sources, and provides accurate decision-making basis for graded treatment.

[0013] Furthermore, the step of obtaining the target treatment plan based on the target monitoring data, the karst fissure water risk level, the karst fissure water characteristic contribution ranking, the pressure field, the flow velocity field, and the structural strain field includes: When the risk level is Level 1 stable, a target treatment plan is obtained based on the flow field and the target monitoring data. When the risk level is a level 2 fluctuation type, the target treatment plan is obtained based on the ranking of the contribution of karst fissure water characteristics and the target monitoring data; When the risk level is Level 3 active, the target treatment plan is obtained based on the pressure field, the flow velocity field, and the structural strain field. When the risk level is a Level IV sudden risk type, a target treatment plan is obtained based on the pressure field, the target monitoring data, and the structural strain field.

[0014] In the above scheme, differentiated input data (flow velocity field and target monitoring data, feature contribution ranking and target monitoring data, pressure field and flow velocity field and structural strain field, pressure field and target monitoring data and structural strain field) are used to generate corresponding target treatment schemes for different risk levels (Level 1 stable, Level 2 fluctuating, Level 3 active, and Level 4 sudden risk). This achieves hierarchical and precise intervention from local blockade to emergency response, avoiding resource waste or safety risks caused by uniform treatment.

[0015] Furthermore, when the risk level is Level 1 stable, the target treatment plan is obtained based on the flow field and the target monitoring data, including: Based on the target monitoring data, seepage flow data and background seepage flow are obtained; The location of isolated fractures is determined based on the seepage flow data and the background seepage flow, and the fracture parameters of the isolated fractures are obtained. The target sealing material and the amount of sealing material used are obtained based on the fracture parameters, the seepage flow data, and the velocity field. The target treatment plan is determined based on the target sealing material and the amount of sealing material used.

[0016] In the above scheme, under the first-level stable risk, the location of isolated cracks is determined based on seepage flow data and background seepage flow, crack parameters are obtained, and the type and amount of sealing material are estimated by combining the flow velocity field. This achieves automatic identification and precise local sealing of low-risk leakage points, minimizing treatment costs and disturbance to the structure while ensuring safety.

[0017] Furthermore, when the risk level is a level two fluctuation type, a target treatment plan is obtained based on the ranking of the contribution of karst fissure water characteristics and the target monitoring data, including: The target monitoring point area is obtained by ranking the contribution of the karst fissure water characteristics. Based on the target monitoring data, the target monitoring infiltration flow rate in the target monitoring point area is obtained; Based on the target monitored seepage flow rate, the grouting pressure and grouting volume are obtained, and then the target treatment plan is determined based on the grouting pressure and grouting volume.

[0018] In the above scheme, under the secondary fluctuation risk, the target monitoring point area is located by ranking the contribution of karst fissure water characteristics. Based on the seepage flow data of this area, the grouting pressure and grouting volume are calculated, realizing low-pressure limited grouting for fluctuation seepage. This not only inhibits the development of seepage but also avoids damage to the structure caused by excessive grouting, thus improving the economy and controllability of the treatment.

[0019] Another embodiment of the present invention provides a karst fissure water treatment device, comprising: The basic data acquisition module is used to acquire target monitoring data and a set of points to be interpolated in the construction-affected area. A water pressure strain data acquisition module is used to obtain water pressure data and strain data based on the target monitoring data; The pressure and velocity field acquisition module is used to obtain the pressure field based on the set of interpolation points, the preset water pressure measurement points and the water pressure data, and to obtain the velocity field based on the pressure field and the preset water parameters. The structural strain field acquisition module is used to obtain the structural strain field based on the set of interpolation points, preset strain measurement points and the strain data; The karst fissure water feature vector acquisition module is used to obtain the karst fissure water feature vector based on the target monitoring data, pressure field, flow velocity field and structural strain field; The contribution ranking acquisition module is used to input the target monitoring data, karst fissure water feature vector, pressure field, flow velocity field and structural strain field into the pre-acquired karst fissure water risk model to obtain the karst fissure water risk level and the karst fissure water feature contribution ranking. The target treatment plan acquisition module is used to obtain the target treatment plan based on the target monitoring data, the risk level of karst fissure water, the ranking of the characteristic contribution of karst fissure water, the pressure field, the flow velocity field, and the structural strain field. The treatment module is used to treat the construction-affected area based on the target treatment plan.

[0020] Another embodiment of the present invention provides a terminal device, including: a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, it implements the steps of the karst fissure water treatment method of the present invention.

[0021] Another embodiment of the present invention also provides a computer-readable storage medium item, including: a stored computer program, which, when the computer program is running, controls the device where the computer-readable storage medium is located to perform the steps of a karst fissure water treatment method of the present invention. Attached Figure Description

[0022] To more clearly illustrate the technical solution of this application, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0023] Figure 1 This is a schematic flowchart of a method for treating karst fissure water according to an embodiment of the present invention; Figure 2 This is a schematic diagram of a karst fissure water treatment device provided in an embodiment of the present invention. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0026] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly defined.

[0027] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0028] In the description of the embodiments in this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0029] In the description of the embodiments of this application, the term "multiple" refers to two or more (including two), similarly, "multiple sets" refers to two or more (including two sets), and "multiple pieces" refers to two or more (including two pieces).

[0030] In the description of the embodiments of this application, unless otherwise expressly specified and limited, technical terms such as "installation," "connection," "joining," and "fixing" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. For those skilled in the art, the specific meaning of the above terms in the embodiments of this application can be understood according to the specific circumstances.

[0031] See Figure 1 To address the aforementioned problems in the prior art and improve the reliability of karst fissure water treatment, an embodiment of the present invention provides a method for treating karst fissure water, comprising: Step S1: Obtain target monitoring data and the set of points to be interpolated in the construction impact area; Step S2: Obtain water pressure data and strain data based on the target monitoring data; Step S3: Obtain the pressure field based on the set of interpolation points, the preset water pressure measurement points and the water pressure data, and obtain the flow velocity field based on the pressure field and the preset water parameters; Step S4: Obtain the structural strain field based on the set of interpolation points, the preset strain measurement points, and the strain data; Step S5: Obtain the karst fissure water characteristic vector based on the target monitoring data, pressure field, flow velocity field, and structural strain field; Step S6: Input the target monitoring data, karst fissure water feature vector, pressure field, flow velocity field and structural strain field into the pre-acquired karst fissure water risk model to obtain the karst fissure water risk level and the ranking of karst fissure water feature contribution. Step S7: Based on the target monitoring data, the risk level of karst fissure water, the ranking of the contribution of karst fissure water characteristics, the pressure field, the flow velocity field, and the structural strain field, obtain the target treatment plan; Step S8: Treat the construction-affected area based on the target treatment plan.

[0032] In the above scheme, by acquiring target monitoring data and a set of interpolation points in the construction-affected area, the pressure field, flow velocity field, and structural strain field are reconstructed, and the discrete measurement point information is reconstructed into a continuous spatial field. This overcomes the defect that limited measurement points cannot reflect the overall spatial distribution, providing a spatial quantitative basis for determining the subsequent target treatment scheme, thus obtaining a more accurate and reliable target treatment scheme. Then, based on this, the karst fissure water feature vector is extracted, and the target monitoring data, feature vector, and reconstructed field are jointly input into the pre-acquired karst fissure water risk model to obtain the accurate karst fissure water risk level and the ranking of karst fissure water feature contribution. This realizes intelligent classification and risk tracing under multi-source information fusion. Then, based on the target monitoring data, the karst fissure water risk level, the ranking of karst fissure water feature contribution, the pressure field, flow velocity field, and structural strain field, the target treatment scheme is obtained. Based on the target treatment scheme, the construction-affected area is treated, thereby effectively improving the reliability of karst fissure water treatment.

[0033] In another embodiment, the step of acquiring the target monitoring data in the construction impact area and the set of interpolation points includes: Obtain the original monitoring data of the construction-affected area; The original monitoring data is subjected to outlier detection based on a preset outlier detection algorithm to obtain the first intermediate monitoring data. The first intermediate monitoring data is imputed for missing values ​​based on a preset missing value imputation algorithm to obtain the second intermediate monitoring data. The target monitoring data is obtained based on the preset dynamic time warping algorithm and the second intermediate monitoring data.

[0034] It should be noted that during the cut-and-cover reverse construction process, considering the characteristics of the retaining structure, the base slab structure, and the reverse construction stage, hydrogeological pre-exploration is first conducted. A small number of pilot boreholes are drilled at pre-determined key locations within the construction impact area (such as the outside of the retaining structure and below the base slab). Pumping and tracing tests are performed in these pilot boreholes to preliminarily determine the main water-conducting channels, flow direction, flow velocity, and permeability of the karst fissure water, among other key hydrogeological parameters. Subsequently, sensing units are deployed, and each sensing unit includes at least one or more of the following: (1) For the monitoring unit of fissure water pressure and seepage flow: it should be deployed in the main water channel and its influence area revealed by the pre-detection in order to capture the most sensitive hydraulic change signal; the monitoring unit can adopt sensors such as existing vibrating wire piezometers and Doppler ultrasonic flow meters.

[0035] (2) For turbidity / sand content monitoring units: they should be mainly deployed in the downstream direction of the main channel and in key structural nodes (such as construction joints and expansion joints) in order to detect signs of seepage damage (such as piping) as early as possible; the monitoring units can use existing equipment such as scattering infrared turbidity sensors, plastic fiber optic turbidity sensors or multi-parameter water quality analyzers.

[0036] (3) For the grouting pressure feedback unit: it is installed in the area where grouting is planned and integrated with the grouting hole; its grouting pressure feedback unit can use a flat diaphragm grouting pressure sensor, diaphragm pressure transmitter or strain gauge grouting special pressure gauge and other special pressure sensors with anti-clogging and wear-resistant characteristics.

[0037] (4) For structural deformation / strain monitoring units: densely deploy them in structurally sensitive areas (such as the connection between the base plate and the enclosure structure) in areas with active fissure water; the monitoring units can be strain / stress sensors with long-term stability and waterproof and pressure-resistant characteristics, such as vibrating wire strain gauges, fiber optic strain sensors, resistance strain gauges or steel stress gauges.

[0038] The deployment plan takes into account the characteristics of cut-and-cover reverse construction procedures (such as excavation sequence and support removal time) and limited space, and activates the sensing units in stages to ensure that the monitoring system can operate continuously and effectively in complex construction environments.

[0039] During the cut-and-cover reverse construction process, based on the optimized deployment of sensing units, an adaptive sampling strategy and pre-installed edge computing server nodes in the construction impact area are used to collect, control, spatiotemporally fuse, and extract features from multi-source monitoring data in real time, providing high-quality, high-dimensional feature vectors for subsequent intelligent risk assessment. Specifically: (1) Perform adaptive multi-source data acquisition to obtain the original monitoring data of the construction-affected area: To address the characteristics of different types of sensing units and changes in construction stages, the system employs a dynamic sampling rate adjustment mechanism: Initially, the system typically enters a routine monitoring phase, acquiring slowly varying signals such as water pressure and seepage flow at a base frequency (e.g., 1 Hz). When the rate of change of any monitored parameter exceeds a preset threshold (e.g., water pressure abrupt change rate > 0.5 kPa / min) or construction enters a critical phase (e.g., excavation exposes a new free face), the system automatically increases the sampling frequency of the relevant sensors to above 10 Hz and simultaneously activates auxiliary sensors in the adjacent area (e.g., turbidity meters, strain gauges) to form focused, intensive observation. Active tracer triggering is also employed, involving the injection of the environmentally friendly tracer sodium fluorescein and the activation of high-precision optical sensors for dynamic tracer monitoring to obtain information on fracture connectivity and flow velocity fields. Based on these operations, the original monitoring data for the construction-affected area is obtained.

[0040] (2) Then, quality control and anomaly repair were performed on the original monitoring data to obtain the target monitoring data: The collected raw monitoring data is processed in real time by the edge computing server node, and the following quality control operations are performed: The first step is outlier detection. A preset outlier detection algorithm is used to identify outliers caused by sensor malfunctions, electromagnetic interference, etc., and remove them to obtain the first intermediate monitoring data. The preset outlier detection algorithm can be an existing density-based clustering method (DBSCAN) or an isolated forest algorithm. Next, missing value imputation is performed. For brief data gaps in the first intermediate monitoring data, a preset missing value imputation algorithm is used. Specifically, the preset missing value imputation algorithm uses the spatiotemporal k-nearest neighbor algorithm, combined with neighboring sensors and historical data from the same period, for imputation. If the missing time is longer, a data repair model based on generative adversarial networks (GANs) is activated to generate alternative data that conforms to physical laws. Based on the above processing, the second intermediate monitoring data is obtained.

[0041] Furthermore, the confidence level of the second intermediate monitoring data can be assessed, and a confidence level label (such as high, medium, or low) can be attached to each data point for reference by the subsequent fusion model. Understandably, the confidence level can be determined based on the sensor's own condition, such as based on sensor self-diagnostic information (battery level, signal strength, whether it is within the calibration validity period, etc.), with higher confidence levels for higher battery level, stronger signal strength, and being within the calibration validity period. It can also be determined based on data consistency, such as a lower degree of deviation between the current data and historical baselines or adjacent sensor data, resulting in higher confidence. Alternatively, it can be determined based on anomaly detection results; for example, if a data point is marked as a suspected outlier by the Isolated Forest or DBSCAN algorithm but not removed (e.g., only near the boundary), its confidence level will be reduced.

[0042] (3) Then, spatiotemporal alignment is performed to obtain target monitoring data: Because different sensors differ in sampling time, spatial location, and physical units, rigorous spatiotemporal registration and fusion are required. The core of this step lies in reconstructing a continuous physical field based on a finite number of discrete measurement points; therefore, spatiotemporal alignment is essential, specifically including: The existing preset Dynamic Time Warping (DTW) algorithm is used to synchronize the time series data collected by each sensor to a unified time base. DTW eliminates time misalignment caused by signal propagation delay, differences in sensor response speed, or inconsistent sampling start times by calculating the optimal nonlinear mapping between two time series. For example, for sensor sequences A and B, the DTW algorithm solves for a curved path that minimizes the cumulative distance, thereby aligning the two sequences on the time axis.

[0043] Furthermore, for steps S2-S3: "Obtain water pressure data and strain data based on the target monitoring data; obtain a pressure field based on the set of interpolation points, preset water pressure measurement points, and the water pressure data; and obtain a flow velocity field based on the pressure field and preset water parameters," specifically: The target monitoring data includes various data collected and processed by various monitoring units (sensors), including water pressure data.

[0044] Based on the measured values ​​(i.e., water pressure data) from a finite number of discrete measuring points (such as N preset water pressure measuring points), the hydraulic parameter field of the entire construction area (three-dimensional space Ω) is reconstructed using Kriging interpolation or radial basis function neural network (RBFNN), including the pressure field P(x,y,z,t) and the velocity field. (x, y, z, t), where x, y, z represent the coordinates in the x, y, and z directions, and t is time. The specific implementation is as follows: Based on the expected value of the spatial distance h between all preset water pressure measuring points and the square of the difference in water pressure data, the existing theoretical variogram model (such as the spherical model, exponential model, or Gaussian model) is fitted to obtain the variogram γ().

[0045] For any point in the set of points to be interpolated Using the known measured values ​​(i.e., water pressure data) P( from preset water pressure measuring points) By solving the Kriging equations: ; ; in, Here, is the weighting coefficient, μ is the preset Lagrange multiplier, and n is equal to N, representing the different indices of the N preset water pressure measuring points. For the i-th preset water pressure measuring point, For the j-th preset water pressure measuring point, i and j take values ​​from 1 to N. The point to be interpolated. It uses a variogram to describe two points. and Spatial correlation or difference between them. The closer the distance, the smaller the variogram value (stronger the correlation); the farther the distance, the larger the value (weaker the correlation). It is usually obtained by fitting experimental data. It also uses a variogram to describe the known preset water pressure measurement points. Interpolation point The variogram values ​​between the two reflect their spatial correlation. After solving for the weighting coefficients using the above formula, Interpolation results at for: ; in, Preset water pressure measurement points Water pressure data at the location.

[0046] The construction-affected area is divided into regular grids, and the interpolation of the interpolation points in the above set of interpolation points is performed on each grid point to obtain the continuous pressure field P(x,y,z,t).

[0047] Then, the flow velocity field is obtained based on the pressure field and the preset water parameters. , Where K is the permeability coefficient, obtained from pre-detection or field tests; μw is the dynamic viscosity of water; the preset water parameters include the permeability coefficient and dynamic viscosity of water mentioned above; The pressure gradient is obtained from the pressure field. The specific acquisition process is described below for calculating the pressure gradient magnitude. Adding the pre-obtained direction information to the pressure gradient magnitude gives the pressure gradient. The flow velocity field is then calculated using the pressure gradient from the pressure field. (x,y,z,t).

[0048] Further, regarding step S4: obtaining the structural strain field based on the set of interpolation points, the preset strain measurement points, and the strain data, specifically: The target monitoring data includes various data collected and processed by various monitoring units (sensors), including strain data.

[0049] Based on the strain data from M preset water pressure measuring points, the structural strain field of the entire construction area (three-dimensional space Ω) is reconstructed. The process is similar to the process of obtaining the pressure field described above, and the structural strain field is finally obtained.

[0050] In another embodiment, obtaining the karst fissure water feature vector based on the target monitoring data, pressure field, flow velocity field, and structural strain field includes: Based on the target monitoring data, a subset of the measurement point signal features is obtained; A subset of spatial gradient features is obtained based on the pressure field and flow velocity field; Based on the structural strain field and the target monitoring data, a multi-source coupling feature subset is obtained; The feature vector of karst fissure water is obtained by splicing together the feature subsets of the measuring point signals, the spatial gradient feature subsets, and the multi-source coupling feature subsets.

[0051] It should be noted that, to facilitate efficient risk assessment in subsequent steps, a series of high-order features are extracted from the fused data to form a feature vector characterizing the current state of karst fissure water. The acquisition of this karst fissure water feature vector specifically involves: (1) Time domain characteristics: Obtain the mean, variance, rate of change, acceleration, peak factor, etc. of various data in the target monitoring data to form time domain characteristics.

[0052] (2) Frequency domain features: Through fast Fourier transform (FFT) or wavelet packet decomposition, the main frequency, energy spectrum, wavelet entropy, etc. of water pressure and flow rate are extracted from the target monitoring data to form frequency domain features, so as to identify the seepage pulsation pattern.

[0053] (3) Nonlinear characteristics: Obtain the seepage flow time series from the target monitoring data, calculate the fractal dimension and Lyapunov exponent of the seepage flow time series, and determine whether the system has entered a chaotic state (a precursor to sudden water inrush). The specific calculation method is as follows: (A) Steps for calculating fractal dimension (box dimension): Take a seepage flow time series of length N {Q( ),Q( ),...,Q( )},Q( () represents the i-th seepage flow rate; Normalize the seepage flow time series to the [0,1] interval; For scale ε= (k=1,2,...,log2N), cover the trajectory curve of the seepage flow time series with a grid of side length ε, and count the number of grid points containing the curve N(ε); In a double logarithmic coordinate system, a linear fit is performed with log(1 / ε) as the abscissa and logN(ε) as the ordinate. The slope of the fitted line is the box dimension Dbox. The closer the Dbox is to 1, the more regular the seepage flow time series (laminar flow); the closer the Dbox is to 1.5~2, the more complex the seepage flow time series (a precursor to turbulence or chaos). When the Dbox continues to rise and exceeds the threshold (e.g., 1.7), it indicates a tendency towards instability.

[0054] (B) Steps for calculating the Lyapunov index: The Wolf reconstruction method is employed: First, the phase space of the seepage flow time series is reconstructed, taking the embedding dimension m (determined by the pseudo-nearest neighbor method) and the time delay τ (determined by the mutual information method) to obtain the phase space trajectory { , ,..., }, Let i be the i-th phase space point; Find each phase space point nearest neighbor phase space point Ensure that j(i)≠i and |j(i)−i|>the preset average period; Calculate the distance di(0) between adjacent pairs of points in phase space = || - ||; After evolving by a time step Δt, calculate the new distance di(Δt) = || - ||; Maximum Lyapunov index Estimated by the following formula: ; Where M is the number of phase space points involved in the calculation (i.e., the number of nearest neighbor phase space point pairs that were successfully tracked and used in the calculation).

[0055] when When the value is >0 and continues to increase, it indicates the entry into a chaotic state, with highly irregular seepage behavior, which is a precursor to a sudden surge of water; when When it is close to 0, the current period is either cyclical or quasi-cyclical; when... A value less than 0 indicates stability.

[0056] It should be noted that the feature subset of the measurement point signal includes time-domain features, frequency-domain features, and nonlinear features.

[0057] (4) The spatial gradient feature subset: The spatial gradient feature subset includes the maximum value of the pressure gradient modulus and its location coordinates, the curl of the velocity field, etc. The specific calculation process is as follows: (A) Calculation of pressure gradient modulus: The continuous pressure field P(x,y,z,t) obtained from the reconstruction is at each grid point ( The three components of the pressure gradient are calculated on the surface. , and ),in Let i be the coordinates of the i-th grid point in the x-direction. Let j be the coordinates of the j-th grid point in the y-direction. Let be the coordinates of the k-th grid point in the z-direction, i be the grid index in the x-direction, ranging from 1 to Nx, where Nx is the total number of grid points in the x-direction. j be the grid index in the y-direction, ranging from 1 to Ny, where Ny is the total number of grid points in the y-direction. k be the grid index in the z-direction, ranging from 1 to Nz, where Nz is the total number of grid points in the z-direction. ; ; ; in, For grid points The pressure at the point, taking the value x= y= z= The pressure field P(x,y,z,t) at time; The grid spacing in the x-direction (the distance between two adjacent grid points in the x-direction); y represents the grid spacing in the y direction (the distance between two adjacent grid points in the y direction); z is the grid spacing in the z direction (the distance between two adjacent grid points in the z direction); This is the partial derivative of the pressure in the x-direction (the component of the pressure gradient in the x-direction). This is the partial derivative of the pressure in the y-direction (the component of the pressure gradient in the y-direction). Let be the partial derivative of the pressure in the z-direction (the component of the pressure gradient in the z-direction); that is, ∇P(x,y,z)=( , , It's understandable that 't' was omitted for simplicity, since gradient calculation doesn't involve time differences.

[0058] Then calculate the pressure gradient modulus : ; Traverse all grid points and find The maximum value is recorded, and its coordinates are recorded. ).

[0059] (B) Calculation of the curl of the velocity field: From the reconstructed velocity field (x,y,z,t)=( , , ),in, Let x be the component of the flow velocity in the x-direction. Let be the component of the flow velocity in the y-direction. Let be the component of the flow velocity in the z-direction, and be the curl. Defined as: ( ); The partial derivatives are calculated using the central difference approximation to obtain the three components of the curl. The magnitude of the curl reflects the intensity of the seepage vortex; regions with large curl indicate the presence of flow around or backflow, representing potential unsteady seepage zones.

[0060] (5) Multi-source coupling feature subset: The multi-source coupling feature subset includes the following coupling features: the cross-correlation peak value of the water pressure change rate and the strain change rate; the time difference between the moment of sudden change in turbidity and the moment of sudden drop in water pressure at a nearby monitoring point; and other features used to reveal the intensity of the interaction between seepage and structure. Specifically, water pressure data is obtained from the target monitoring data, and the water pressure change rate is obtained based on the water pressure data; strain data is obtained from the target monitoring data, and the strain change rate is obtained based on the strain data. The water pressure change rate sequence is... With strain change rate sequence Perform cross-correlation analysis and calculate the cross-correlation function. The maximum value of this function is taken as the cross-correlation peak value. Characterize the seepage-structure coupling strength; simultaneously record the time shift corresponding to the peak value. , which characterizes the optimal hysteresis time between water pressure change and structural response.

[0061] The feature vector of karst fissure water is obtained by splicing together the feature subsets of the measuring point signals, the spatial gradient feature subsets, and the multi-source coupling feature subsets.

[0062] Furthermore, the data processing flow of steps S1-S5 above is mainly deployed on edge computing servers near the construction impact area to ensure real-time performance (latency <1 second). The extracted feature vectors and condensed information (rather than the original massive data) are periodically uploaded to the cloud platform for long-term trend analysis and model retraining, forming an edge-cloud collaborative intelligent perception architecture.

[0063] In another embodiment, the karst fissure water risk model includes a spatial attention module, a temporal attention module, and a multi-scale feature fusion layer. The step of inputting the target monitoring data, karst fissure water feature vectors, pressure field, flow velocity field, and structural strain field into the pre-acquired karst fissure water risk model to obtain the karst fissure water risk level and the ranking of karst fissure water feature contributions includes: The target monitoring data, karst fissure water feature vector, pressure field, flow velocity field, and structural strain field are input into the karst fissure water risk model. This allows the spatial attention module to obtain spatial attention weighted field features and spatial contribution ranking based on a preset graph attention network, the pressure field, flow velocity field, and structural strain field. The temporal attention module then obtains temporal attention weighted sequence features and temporal contribution ranking based on the target monitoring data. Finally, the multi-scale feature fusion layer nonlinearly fuses the karst fissure water feature vector, the spatial attention weighted field features, and the temporal attention weighted sequence features to obtain a comprehensive risk index. The risk level of the karst fissure water is obtained based on the comprehensive risk index, and the characteristic contribution ranking of the karst fissure water is obtained based on the spatial contribution ranking and the temporal contribution ranking.

[0064] It should be noted that the karst fissure water risk model employs an existing multi-scale spatiotemporal attention network. The karst fissure water risk model includes a spatial attention module, a temporal attention module, and a multi-scale feature fusion layer. The target monitoring data, karst fissure water feature vectors, pressure field, flow velocity field, and structural strain field are input into the karst fissure water risk model to obtain the karst fissure water risk level and the ranking of karst fissure water feature contributions. The specific process is as follows: The karst fissure water risk model comprises a spatial attention module, a temporal attention module, and a multi-scale feature fusion layer. The spatial attention module takes the pressure field, velocity field, and structural strain field as input. This module employs a pre-defined graph attention network (GAT) to calculate the risk weight of each spatial node (i.e., monitoring point or grid point) to the overall seepage risk based on the input pressure field, velocity field, and structural strain field. Specifically, the graph attention network calculates the Euclidean distance between nodes based on their spatial coordinates and constructs an adjacency matrix (nodes with a distance less than a preset threshold are considered neighbors). Then, using the graph attention mechanism, it aggregates features such as pressure gradient, velocity direction, and structural strain rate of neighboring nodes, assigning an attention coefficient to each node (representing the node's contribution to the overall seepage risk). Through aggregation and updating of the graph attention network, the spatial attention weighted field features are output, along with the spatial contribution ranking of each spatial node (i.e., the order of contribution of each location to the current risk).

[0065] The time attention module takes historical time-series data (including time series of monitoring parameters such as water pressure, seepage flow, and turbidity) from the target monitoring data as input. This module employs the Transformer time encoding mechanism and multi-head self-attention to process the input time series, calculating an attention weight for each historical moment. This weight represents the contribution of that moment to the current comprehensive risk index prediction. The attention weights for each historical moment are sorted from largest to smallest, thus obtaining the time contribution ranking. Simultaneously, these attention weights are used to weighted aggregate the sequence features, outputting time attention-weighted sequence features.

[0066] Subsequently, based on the spatial contribution ranking, the spatial dimension risk tracing result (i.e., the order of contribution of each spatial location to the current risk) is obtained, and based on the temporal contribution ranking, the temporal dimension risk tracing result (i.e., the order of contribution of each historical moment to the current risk) is obtained. The spatial contribution ranking is used to guide the determination of grouting hole locations and reinforcement locations, and the temporal contribution ranking is used to determine the risk evolution type (sudden or cumulative). Both serve as the basis for subsequent graded treatment decisions, obtaining the comprehensive contribution value of this feature, and then sorting by value to obtain the feature contribution ranking.

[0067] Meanwhile, the multi-scale feature fusion layer splices together the karst fissure water feature vector, the spatial attention weighted field feature output by the spatial attention module, and the temporal attention weighted sequence feature output by the temporal attention module.

[0068] The fusion layer then uses a multilayer perceptron (MLP) to perform nonlinear fusion on the stitched features, and finally outputs a scalar value, namely the comprehensive risk index.

[0069] The comprehensive risk index is compared with multiple preset thresholds to classify karst fissure water risk levels into four levels. It is understood that these preset thresholds can be dynamically adjusted using existing Bayesian online variable point detection algorithms, or they can be manually preset. The specific karst fissure water risk levels are as follows: Level 1 Stable: The overall risk index is below the lower threshold. Level 2 Volatile: The overall risk index is between the lower and middle thresholds. Level 3 Active: The overall risk index is between the middle and upper thresholds. Level 4 Sudden Risk: The overall risk index is above the upper threshold.

[0070] Meanwhile, based on the above spatial contribution ranking and temporal contribution ranking, the feature contribution ranking of each feature in the karst fissure water feature vector is derived, that is, the order of importance of each karst fissure water feature vector to the current risk, which is used for risk tracing and graded treatment in subsequent steps.

[0071] In another embodiment, the step of obtaining the target treatment plan based on the target monitoring data, the karst fissure water risk level, the karst fissure water characteristic contribution ranking, the pressure field, the flow velocity field, and the structural strain field includes: When the risk level is Level 1 stable, a target treatment plan is obtained based on the flow field and the target monitoring data. When the risk level is a level 2 fluctuation type, the target treatment plan is obtained based on the ranking of the contribution of karst fissure water characteristics and the target monitoring data; When the risk level is Level 3 active, the target treatment plan is obtained based on the pressure field, the flow velocity field, and the structural strain field. When the risk level is a Level IV sudden risk type, a target treatment plan is obtained based on the pressure field, the target monitoring data, and the structural strain field.

[0072] In another embodiment, when the risk level is Level 1 stable, obtaining a target treatment plan based on the flow field and the target monitoring data includes: Based on the target monitoring data, seepage flow data and background seepage flow are obtained; The location of isolated fractures is determined based on the seepage flow data and the background seepage flow, and the fracture parameters of the isolated fractures are obtained. The target sealing material and the amount of sealing material used are obtained based on the fracture parameters, the seepage flow data, and the velocity field. The target treatment plan is determined based on the specified sealing material and the amount of sealing material used.

[0073] It should be noted that Level 1 stability refers to the comprehensive risk index. Common measures include routine monitoring and localized sealing. Specifically, seepage flow data is obtained based on the target monitoring data. That is, the seepage flow data Q collected by the seepage flow monitoring unit and the background seepage flow Q0 are extracted from the target monitoring data. It is understood that when the seepage flow data includes a time dimension, it becomes a seepage flow time series, which is a common form of seepage flow data. When the seepage flow data Q near an isolated fracture exceeds a preset multiple (e.g., 1.5 times) of the background seepage flow, the location of the isolated fracture is marked, and the seepage flow data near that isolated fracture is used as... Obtain the fracture parameters corresponding to the isolated fracture location, including: fracture width W (which can be measured on-site or obtained from experience), seepage path length L, and head difference Δh. These parameters are collectively referred to as fracture parameters. Then, based on the fracture parameters, the seepage flow data, and the velocity field, obtain the target sealing material and the amount of sealing material used. Specifically: (1) First, estimate the equivalent hydraulic opening b of the fracture: For isolated fractures, according to the cubic law, the seepage flow data of that isolated fracture is used. Using the velocity estimate v obtained from the velocity field, the equivalent hydraulic opening b is calculated: ; Where ρ is the density of water, g is the acceleration due to gravity, and μw is the dynamic viscosity. The equivalent hydraulic opening b can be solved using the above formula.

[0074] In another embodiment, when the risk level is a level two fluctuation type, a target treatment plan is obtained based on the ranking of the contribution of karst fissure water characteristics and the target monitoring data, including: The target monitoring point area is obtained by ranking the contribution of the karst fissure water characteristics. Based on the target monitoring data, the target monitoring infiltration flow rate in the target monitoring point area is obtained; Based on the target monitored seepage flow rate, the grouting pressure and grouting volume are obtained, and then the target treatment plan is determined based on the grouting pressure and grouting volume.

[0075] (2) Select the type of sealing material: Based on the estimated velocity v and equivalent hydraulic opening b at the isolated fracture in the velocity field, the following rules are used to recommend the sealing material to obtain the target sealing material: If v < 0.1 m / s and b < 1 mm, cement-based ultrafine slurry is recommended; If 0.1≤v<0.5 m / s and 1≤b<5 mm, cement-water glass two-component grout is recommended; If v ≥ 0.5 m / s or b ≥ 5 mm, chemical grout (such as polyurethane) or quick-setting cement is recommended.

[0076] (3) Estimate the amount of sealing material needed: Calculate the fracture volume =b·W·L, multiplied by the filling factor η (preferably 1.2~2.0), that is: ; in, The amount of sealing material used.

[0077] Finally, the target material type obtained above and the calculated amount of sealing material are used as a local sealing scheme and output as the target treatment scheme.

[0078] In another embodiment, when the risk level is a level two fluctuation type, a target treatment plan is obtained based on the ranking of the contribution of karst fissure water characteristics and the target monitoring data, including: The target monitoring point area is obtained by ranking the contribution of the karst fissure water characteristics. Based on the target monitoring data, the target monitoring infiltration flow rate in the target monitoring point area is obtained; Based on the target monitored seepage flow rate, the grouting pressure and grouting volume are obtained, and then the target treatment plan is determined based on the grouting pressure and grouting volume.

[0079] It should be noted that the second-order fluctuation type, i.e., the lower threshold... Overall Risk Index Middle threshold The measures adopted were low-pressure, limited-volume grouting, specifically: The target monitoring point area is obtained based on the ranking of the contribution of karst fissure water characteristics: From the ranking of the contribution of karst fissure water characteristics, the monitoring parameter (e.g., the rate of change of seepage flow) corresponding to the characteristic with the highest contribution is identified. The location of the monitoring point with the largest contribution of this parameter and its adjacent area (e.g., within a radius of 1-2 m) are determined as the grouting target area, which is used as the target monitoring point area.

[0080] Then, based on the target monitoring data, the target monitoring seepage flow rate in the target monitoring point area is obtained: from the target monitoring data, the target monitoring seepage flow rate Q(t) (i.e., the seepage flow rate time series) collected by the seepage flow monitoring unit in the target monitoring point area is extracted. This data is in time series form and is used for subsequent calculation of the rate of change and cumulative amount.

[0081] Then, based on the target monitored seepage flow rate, the grouting pressure and grouting volume are obtained, and the target treatment plan is determined based on the grouting pressure and grouting volume, specifically: The target monitoring point area with the largest contribution to the rate of change in seepage flow is selected in the risk tracing process; The grouting pressure is obtained using the following formula. : ; in, The initial grouting pressure (preset based on experience or formation conditions). t is an adjustable coefficient, and t is the grouting duration; This represents the maximum rate of change in seepage.

[0082] Determine the target time period for the blockade. , ] Calculate the grouting volume during this period. : ; in, This is a preset safety factor.

[0083] Finally, the grouting pressure Pg and grouting volume were... As the core parameter of the low-pressure limited grouting scheme, combined with the grouting hole location (i.e. the target monitoring point area), the output is a two-level fluctuating target treatment scheme.

[0084] Furthermore, when the risk level is Level 3 active, a target treatment plan is obtained based on the pressure field, the flow velocity field, and the structural strain field, specifically: Level 3 active type, i.e., medium threshold Overall Risk Index Upper threshold The measures typically involve dynamic targeted grouting and adaptive structural reinforcement, specifically: First, the core region of the main seepage channel is identified, based on the reconstructed velocity field. The set of grid points that simultaneously satisfy the following two conditions is determined to be the core region of the main seepage channel: the magnitude of the flow velocity is greater than the global preset quantile (e.g., the 85th quantile); the pressure gradient divergence satisfies ∇(∇P| ||∇P||)<0 (i.e., the pressure gradient directions converge).

[0085] Next, the grouting hole locations and sequence were optimized, and the core area of ​​the main seepage channel was divided into regular grids. For each grid cell, its average velocity vector was calculated. =( , , The optimal azimuth angle of the grouting hole is determined to ensure that the borehole axis is perpendicular to the main flow direction. Specifically, the unit vector n of the borehole axis... =(n x ,n y ,n z ) Must meet: =0; And ||n ||=1. The solution space of this equation is a circle. To determine a unique solution, a constraint can be added: minimize the angle between the projection of the borehole axis onto the horizontal plane and a certain reference direction (such as the x-axis), or take n. To be perpendicular to And a vector pointing in the negative direction of the pressure gradient (i.e., the direction of incoming water): Where g =−∇P(g (This represents the negative direction of the pressure gradient). This formula gives a direction that is perpendicular to both the flow velocity and the pressure gradient, which is usually the ideal orientation of the grouting hole.

[0086] After obtaining the recommended borehole azimuth angle for each grid, a particle swarm optimization algorithm is used to select the combination of borehole locations that maximizes the ratio of plugging volume to borehole length, under constraints such as construction space and drilling rig accessibility. The fitness function is: ; Where f is the fitness function value, representing the seepage intensity (i.e., plugging efficiency) blocked per unit borehole length. The average flow velocity (scalar) within the k-th grid cell; Let j be the drilling length of the j-th grouting hole. Let j be the grid volume of the k-th grid. j takes values ​​from 1 to p, where p represents the total number of grouting holes.

[0087] Then, based on the reconstructed structural strain field ε(x,y,t), the strain rate field is calculated. Areas whose strain change rate exceeds a preset threshold and overlaps with the high gradient zone of seepage (i.e., the core area of ​​the main seepage channel) are identified as key reinforcement areas. A multi-criteria decision-making process is used to prioritize these key reinforcement areas (for example, strain rate is weighted at 0.6, and distance from the seepage channel is weighted at 0.4), and the reinforcement order is determined from highest to lowest weighted score.

[0088] The optimized grouting hole positions, hole sequence, and priority of key reinforcement parts are used as dynamic targeted grouting and adaptive structural reinforcement schemes, and the output is a three-level active target treatment scheme.

[0089] Furthermore, when the risk level is a Level IV sudden risk type, a target treatment plan is obtained based on the pressure field, the target monitoring data, and the structural strain field, specifically: Level 4 sudden risk type, i.e., comprehensive risk index The commonly used measures are emergency response, specifically: Based on the pressure field, the spatial point corresponding to the maximum pressure gradient modulus is found; at the same time, the turbidity time series is extracted from the target monitoring data to identify the moment when the turbidity first changes abruptly and the corresponding monitoring point location; the spatial midpoint of the above two points is taken as the rapid sealing hole location.

[0090] Pressure or seepage flow time series are extracted from the target monitoring data, and the spectrum is obtained by Fast Fourier Transform (FFT). The frequency with the largest amplitude is taken as the dominant frequency of seepage pulsation. : like >0.5 Hz indicates fast seepage velocity and strong turbulence, and quick-setting materials (such as polyurethane) are recommended. like ≤0.5 Hz indicates relatively stable seepage, and slow-setting high-strength materials (such as cement-water glass) are recommended.

[0091] Based on the pressure field P(x,y,z,t), the pressure gradient field ∇P is calculated. A set of candidate points is selected on the structural boundary of the construction impact area (such as the retaining structure or the edge of the base slab). For each candidate point, the pressure gradient vector is calculated and extended outwards. If the extension direction is consistent with the pre-defined safety passage direction, and the pressure value at that point is lower than the global average pressure, it is marked as a safe pressure relief zone. The preset standby drainage pump in this area is automatically activated.

[0092] Based on the reconstructed structural strain field ε(x,y,t) and its rate of change field (x, y, t) Generate a strain sensitivity map (plotting the spatial distribution of strain change rate on the BIM model using color charts). Identify the three points with the largest strain change rate and automatically deploy temporary supports.

[0093] Finally, the above-mentioned quick-sealing hole locations, sealing materials, pressure relief zone drainage instructions, and temporary support layout scheme are integrated into an emergency response plan, which is output as a target treatment plan for a level four sudden risk type.

[0094] Furthermore, after treating the affected area according to the target treatment plan, the system continuously acquires post-treatment target monitoring data and repeats the following operations to form a dynamic closed-loop control: (1) After the treatment measures are implemented, target monitoring data of the construction-affected area are continuously collected, and the following actions are performed again based on the target monitoring data: reconstructing the pressure field, flow velocity field and structural strain field; extracting the karst fissure water feature vector; and inputting the karst fissure water risk model to obtain the comprehensive risk index and karst fissure water risk level.

[0095] The changes before and after treatment were compared from the following three dimensions: risk index comparison, comparing the decrease in the comprehensive risk index before and after treatment; field state comparison, comparing the spatial distribution differences of pressure field, velocity field and structural strain field before and after treatment (such as changes in indicators such as the maximum value of pressure gradient modulus and velocity curl); and feature comparison, comparing the changing trends of each feature value in the karst fissure water feature vector before and after treatment (such as fractal dimension, Lyapunov exponent, cross-correlation peak value, etc.).

[0096] (2) Based on the effect evaluation results, perform the following three-level optimization: (A) Model layer update: Treating cases with poor results (i.e., cases where the risk level is not reduced or is not reduced to the standard after treatment) as negative samples, and incrementally learning the karst fissure water risk model (including spatial attention module, temporal attention module and multi-scale feature fusion layer) to update the model parameters; (B) Parameter layer correction: Collect the measured values ​​of grouting pressure and the measured values ​​of the response of monitoring holes (water pressure, strain, etc.) during the treatment process, and use these data to correct the response sensitivity matrix for subsequent calculation of grouting pressure and grouting volume; (C) Strategy layer evolution: Successful treatment cases (i.e., cases where the risk level is significantly reduced and the structure is safe and stable after treatment) are patterned, summarized into treatment templates, stored in the knowledge base, and experience reuse is realized.

[0097] (3) Each treatment case is transformed into a knowledge triple (including: risk level, treatment measures, and effect evaluation), stored in the construction risk knowledge graph, and linked with the BIM model. The historical risk points, the treatment measures taken, and the effect evaluation are visualized in three-dimensional space to provide decision support for subsequent construction.

[0098] Through the above methods, a dynamic intelligent closed-loop control process for sensing, identifying, treating, feeding back, and evolving karst fissure water is formed.

[0099] like Figure 2 As shown, based on the above method embodiments, corresponding apparatus embodiments are provided; An embodiment of the present invention provides a karst fissure water treatment device, comprising: The basic data acquisition module is used to acquire target monitoring data and a set of points to be interpolated in the construction-affected area. A water pressure strain data acquisition module is used to obtain water pressure data and strain data based on the target monitoring data; The pressure and velocity field acquisition module is used to obtain the pressure field based on the set of interpolation points, the preset water pressure measurement points and the water pressure data, and to obtain the velocity field based on the pressure field and the preset water parameters. The structural strain field acquisition module is used to obtain the structural strain field based on the set of interpolation points, preset strain measurement points and the strain data; The karst fissure water feature vector acquisition module is used to obtain the karst fissure water feature vector based on the target monitoring data, pressure field, flow velocity field and structural strain field; The contribution ranking acquisition module is used to input the target monitoring data, karst fissure water feature vector, pressure field, flow velocity field and structural strain field into the pre-acquired karst fissure water risk model to obtain the karst fissure water risk level and the karst fissure water feature contribution ranking. The target treatment plan acquisition module is used to obtain the target treatment plan based on the target monitoring data, the risk level of karst fissure water, the ranking of the characteristic contribution of karst fissure water, the pressure field, the flow velocity field, and the structural strain field. The treatment module is used to treat the construction-affected area based on the target treatment plan. It is understood that the above-described device embodiments correspond to the method embodiments of the present invention, and can implement the karst fissure water treatment method provided by any of the above-described method embodiments of the present invention.

[0100] It should be noted that the device embodiments described above are merely illustrative, and some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can specifically be implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0101] Based on the above-described embodiment of a karst fissure water treatment method, another embodiment of the present invention provides a terminal device, which includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, it implements a karst fissure water treatment method according to any embodiment of the present invention.

[0102] For example, in this embodiment, the computer program can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the terminal device.

[0103] The terminal device may be a desktop computer, laptop, handheld computer, or cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0104] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the terminal device, connecting all parts of the terminal device via various interfaces and lines.

[0105] Based on the above-described method embodiments, another embodiment of the present invention provides a computer-readable storage medium including a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to execute the karst fissure water treatment method described in any of the above-described method embodiments of the present invention.

[0106] The modules / units integrated in the device / terminal equipment, if implemented as software functional units and sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0107] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A method for treating karst fissure water, characterized in that, include: Acquire target monitoring data and a set of interpolation points in the construction-affected area; Water pressure data and strain data are obtained based on the target monitoring data; The pressure field is obtained based on the set of interpolation points, the preset water pressure measurement points, and the water pressure data; and the flow velocity field is obtained based on the pressure field and the preset water parameters. The structural strain field is obtained based on the set of interpolation points, the preset strain measurement points, and the strain data. Based on the target monitoring data, pressure field, flow velocity field, and structural strain field, the characteristic vector of karst fissure water is obtained; The target monitoring data, karst fissure water feature vector, pressure field, flow velocity field and structural strain field are input into the pre-acquired karst fissure water risk model to obtain the karst fissure water risk level and the ranking of karst fissure water feature contribution. Based on the target monitoring data, the risk level of karst fissure water, the ranking of the contribution of karst fissure water characteristics, the pressure field, the flow velocity field, and the structural strain field, a target treatment plan is obtained; The construction-affected area is treated based on the target treatment plan.

2. The method for treating karst fissure water according to claim 1, characterized in that, The steps for obtaining the target monitoring data and the set of interpolation points in the construction impact area include: Obtain the original monitoring data of the construction-affected area; The original monitoring data is subjected to outlier detection based on a preset outlier detection algorithm to obtain the first intermediate monitoring data. The first intermediate monitoring data is imputed for missing values ​​based on a preset missing value imputation algorithm to obtain the second intermediate monitoring data. The target monitoring data is obtained based on the preset dynamic time warping algorithm and the second intermediate monitoring data.

3. The method for treating karst fissure water according to claim 1, characterized in that, The method for obtaining the karst fissure water characteristic vector based on the target monitoring data, pressure field, flow velocity field, and structural strain field includes: Based on the target monitoring data, a subset of the measurement point signal features is obtained; A subset of spatial gradient features is obtained based on the pressure field and flow velocity field; Based on the structural strain field and the target monitoring data, a multi-source coupling feature subset is obtained; The feature vector of karst fissure water is obtained by splicing together the feature subsets of the measuring point signals, the spatial gradient feature subsets, and the multi-source coupling feature subsets.

4. The method for treating karst fissure water according to claim 1, characterized in that, The karst fissure water risk model includes a spatial attention module, a temporal attention module, and a multi-scale feature fusion layer. The target monitoring data, karst fissure water feature vectors, pressure field, flow velocity field, and structural strain field are input into the pre-acquired karst fissure water risk model to obtain the karst fissure water risk level and the ranking of karst fissure water feature contributions, including: The target monitoring data, karst fissure water feature vector, pressure field, flow velocity field, and structural strain field are input into the karst fissure water risk model. This allows the spatial attention module to obtain spatial attention weighted field features and spatial contribution ranking based on a preset graph attention network, the pressure field, flow velocity field, and structural strain field. The temporal attention module then obtains temporal attention weighted sequence features and temporal contribution ranking based on the target monitoring data. Finally, the multi-scale feature fusion layer nonlinearly fuses the karst fissure water feature vector, the spatial attention weighted field features, and the temporal attention weighted sequence features to obtain a comprehensive risk index. The risk level of the karst fissure water is obtained based on the comprehensive risk index, and the characteristic contribution ranking of the karst fissure water is obtained based on the spatial contribution ranking and the temporal contribution ranking.

5. The method for treating karst fissure water according to claim 1, characterized in that, The target treatment plan is obtained based on the target monitoring data, the karst fissure water risk level, the karst fissure water characteristic contribution ranking, the pressure field, the flow velocity field, and the structural strain field, including: When the risk level is Level 1 stable, a target treatment plan is obtained based on the flow field and the target monitoring data. When the risk level is a level 2 fluctuation type, the target treatment plan is obtained based on the ranking of the contribution of the karst fissure water characteristics and the target monitoring data; When the risk level is Level 3 active, the target treatment plan is obtained based on the pressure field, the flow velocity field, and the structural strain field. When the risk level is a Level IV sudden risk type, a target treatment plan is obtained based on the pressure field, the target monitoring data, and the structural strain field.

6. A method for treating karst fissure water according to claim 5, characterized in that, When the risk level is Level 1 stable, a target treatment plan is obtained based on the flow field and the target monitoring data, including: Based on the target monitoring data, seepage flow data and background seepage flow are obtained; The location of isolated fractures is determined based on the seepage flow data and the background seepage flow, and the fracture parameters of the isolated fractures are obtained. The target sealing material and the amount of sealing material used are obtained based on the fracture parameters, the seepage flow data, and the velocity field. The target treatment plan is determined based on the target sealing material and the amount of sealing material used.

7. A method for treating karst fissure water according to claim 5, characterized in that, When the risk level is a level 2 fluctuation type, a target treatment plan is obtained based on the ranking of the contribution of karst fissure water characteristics and the target monitoring data, including: The target monitoring point area is obtained by ranking the contribution of the karst fissure water characteristics. Based on the target monitoring data, the target monitoring infiltration flow rate in the target monitoring point area is obtained; Based on the target monitored seepage flow rate, the grouting pressure and grouting volume are obtained, and then the target treatment plan is determined based on the grouting pressure and grouting volume.

8. A device for treating karst fissure water, characterized in that, include: The basic data acquisition module is used to acquire target monitoring data and a set of points to be interpolated in the construction-affected area. A water pressure strain data acquisition module is used to obtain water pressure data and strain data based on the target monitoring data; The pressure and velocity field acquisition module is used to obtain the pressure field based on the set of interpolation points, the preset water pressure measurement points and the water pressure data, and to obtain the velocity field based on the pressure field and the preset water parameters. The structural strain field acquisition module is used to obtain the structural strain field based on the set of interpolation points, preset strain measurement points and the strain data; The karst fissure water feature vector acquisition module is used to obtain the karst fissure water feature vector based on the target monitoring data, pressure field, flow velocity field and structural strain field; The contribution ranking acquisition module is used to input the target monitoring data, karst fissure water feature vector, pressure field, flow velocity field and structural strain field into the pre-acquired karst fissure water risk model to obtain the karst fissure water risk level and the karst fissure water feature contribution ranking. The target treatment plan acquisition module is used to obtain the target treatment plan based on the target monitoring data, the risk level of karst fissure water, the ranking of the characteristic contribution of karst fissure water, the pressure field, the flow velocity field, and the structural strain field. The treatment module is used to treat the construction-affected area based on the target treatment plan.

9. A terminal device, characterized in that, The method includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, it implements a method for treating karst fissure water as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, include: A stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform a method for treating karst fissure water as described in any one of claims 1-7.

Images