Method and device for dynamic prediction of cavern seepage

Abstract

This invention provides a method and apparatus for dynamic prediction of cavern seepage, relating to the field of engineering prediction technology. The method includes: establishing a dynamic evolution model of a fracture network based on underground cavern seepage influencing factor data to dynamically update the fractal characteristic parameters of each fractal fracture element; establishing an anisotropic permeability tensor model based on the underground cavern seepage influencing factor data; constructing a deep physical constraint-based geographically weighted regression model based on the anisotropic permeability tensor model and seepage core parameters; inputting new underground cavern seepage influencing factor data into the deep physical constraint-based geographically weighted regression model; and outputting the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of the total cavern seepage volume as the dynamic prediction result of the cavern seepage volume. This invention can simultaneously consider the multi-scale characteristics of the fracture network, spatiotemporal dynamic evolution, and seepage physical constraints, thereby improving the accuracy, stability, and engineering applicability of seepage volume prediction.

Description

Technical Field

[0001] This invention relates to the field of engineering prediction technology, and in particular to a method and apparatus for dynamic prediction of cavern seepage. Background Technology

[0002] Underground caverns (such as underground powerhouses of hydropower stations, water diversion tunnels, traffic tunnels, underground storage facilities, and mine roadways) commonly face seepage during excavation and operation. The seepage field of the surrounding rock is not only affected by groundwater recharge, hydraulic head boundaries, and the permeability of the soil and rock media, but also significantly controlled by the degree of fracture development, fracture connectivity, and anisotropic characteristics. Furthermore, construction measures such as excavation unloading, blasting disturbance, support, and grouting cause the temporal evolution of the surrounding rock stress field and fracture aperture / connectivity, leading to significant spatiotemporal dynamic changes in seepage channels and seepage volume. Therefore, accurate prediction of seepage volume in underground caverns is a crucial technical foundation for optimizing the layout of drainage systems, providing early warning of water inrush risks, and ensuring safe operation and management of the project, directly impacting construction costs, construction safety, and operational stability.

[0003] In engineering practice, seepage prediction is often based on fitting methods using existing monitoring data or numerical simulation techniques. Conventional data fitting methods, such as statistical fitting and regression analysis, are simple and fast, but suitable for short-term trend prediction. They are insensitive to the complexity of geological conditions, lack mechanistic relationships, and only provide data relationships without understanding the cause of seepage. Numerical simulation techniques have clear mechanisms, but they are complex to model, involve a large workload, and have high computational costs, making it difficult to meet the timely response and prediction needs during construction. Currently, with the development of science and technology, the engineering community is constantly exploring better ways to utilize existing mature basic models and technologies to research methods that meet the needs of cavern seepage prediction. Summary of the Invention

[0004] To overcome the problems existing in related technologies, this invention provides a method and device for dynamic prediction of cavern seepage, which can simultaneously take into account the multi-scale characteristics of fracture networks, spatiotemporal dynamic evolution and seepage physical constraints, so as to improve the accuracy, stability and engineering applicability of seepage prediction.

[0005] According to a first aspect of the present invention, a method for dynamically predicting cavern seepage volume is provided, comprising:

[0006] Obtain data on seepage influencing factors in underground caverns;

[0007] A dynamic evolution model of the fracture network is established based on the data of seepage influencing factors in the underground cavern, so as to dynamically update the fractal characteristic parameters of each fractal fracture element;

[0008] An anisotropic permeability tensor model was established based on the data of seepage influencing factors in the underground caverns.

[0009] A deep physical constraint-based geographically weighted regression model is constructed based on the anisotropic permeability tensor model and the core parameters of seepage.

[0010] The new data on seepage influencing factors in underground caverns are input into the deep physical constraint geographic weighted regression model to obtain the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of seepage volume in the entire cavern.

[0011] The predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of the total seepage volume in the cavern are output as the dynamic prediction result of the cavern seepage volume.

[0012] In some exemplary embodiments of the present invention, based on the foregoing scheme, the underground cavern seepage influencing factor data includes rock mass stress monitoring data;

[0013] A dynamic evolution model of the fracture network is established based on the seepage influencing factor data of the underground cavern, to dynamically update the fractal characteristic parameters of each fractal fracture element, including:

[0014] Extract fracture geometric features from the seepage influencing factor data of the underground cavern;

[0015] The fractal characteristic parameters in the fracture geometry are extracted using the multifractal detrending fluctuation analysis method.

[0016] Based on the fractal characteristic parameters, a clustering algorithm is used to divide the underground cavern into several fractal fracture elements that are homogeneous internally and significantly different externally.

[0017] Based on the rock mass stress monitoring data and combined with the effective stress principle, the change in rock mass stress during construction excavation and support is calculated, and then the change law of fracture aperture is derived, and the fractal characteristic parameters of each fractal fracture element are dynamically updated.

[0018] In some exemplary embodiments of the present invention, based on the foregoing scheme, the underground cavern seepage influencing factor data includes fracture orientation data;

[0019] Based on the data on seepage influencing factors in underground caverns, an anisotropic permeability tensor model is established, including:

[0020] The dominant seepage direction of the underground cavern is determined based on the fracture orientation data;

[0021] Establish a principal direction coordinate system for permeability based on the dominant seepage direction;

[0022] Establish the anisotropic permeability tensor model within the principal directional coordinate system.

[0023] In some exemplary embodiments of the present invention, based on the foregoing scheme, constructing a deep physical constraint-based geographically weighted regression model based on the anisotropic permeability tensor model and seepage core parameters includes:

[0024] Based on the fractal fracture element and the fractal characteristic parameters of each fractal fracture element, a power coupling formula for the permeability coefficient is constructed.

[0025] The anisotropic permeability tensor model is assigned values ​​according to the power coupling formula of the permeability coefficient to obtain the anisotropic permeability tensor of each fractal fracture element.

[0026] The theoretical seepage volume of each fractal fracture element is calculated using the anisotropic permeability tensor of each fractal fracture element and the seepage core parameters.

[0027] The basic spatial weights between different fractal fracture elements are calculated using a Gaussian kernel function. At the same time, construction disturbance factors and geological similarity factors are introduced to dynamically adjust the basic spatial weights to obtain the spatial weights.

[0028] Based on the spatial weights of all fractal fracture elements, a spatial weight matrix is ​​obtained;

[0029] Based on the theoretical seepage volume of each fractal fracture element and the spatial weight matrix, a deep physical constraint-based geographically weighted regression model is established, using a geographically weighted regression model as the foundation.

[0030] In some exemplary embodiments of the present invention, based on the foregoing scheme, the spatial weights include:

[0031]

[0032] in, For the first The fractal fracture element and the first The spatial Euclidean distance between the centers of each fractal fracture element. For bandwidth parameters; The construction disturbance weight is dynamically adjusted based on the distance between the excavation face and the fractal fracture element. The smaller The larger; The geological similarity weight is determined based on the similarity measure of the lithology and / or fracture characteristics of fractal fracture elements; the higher the similarity, the greater the weight. The larger.

[0033] In some exemplary embodiments of the present invention, based on the foregoing scheme, and according to the theoretical seepage volume of each fractal fracture element and the spatial weight matrix, the deep physical constraint-based geographically weighted regression model is established based on the geographically weighted regression model, including:

[0034]

[0035] in, For the first Measured water seepage of each fractal fracture element; For constant terms; For the first The fractal fracture element corresponding to the first One effective influencing factor, The number of effective influence factors; For the first The fractal fracture element corresponding to the first The spatial local regression coefficients are obtained by using the same method as the first... The spatial weight matrix corresponding to each fractal fracture element The weighted least squares solution is obtained, and it satisfies: , For the first The regression coefficient vector of each fractal fracture element For an effective influence factor matrix, This is the measured seepage vector; These are adaptive physical constraint weights used to balance the contribution of data-driven regression fitting and physical constraints, and satisfy the following conditions: , The maximum weight value and , For the weight growth rate, The number of training iterations for the model; For the first The theoretical seepage volume of each fractal fracture element, and satisfying the following conditions: in, For the first The components or equivalent permeability parameters of the anisotropic permeability tensor of a fractal fracture element. For hydraulic gradient, The cross-sectional area of ​​the water passage. The water content is the fractal fracture element.

[0036] In some exemplary embodiments of the present invention, based on the foregoing scheme, before inputting the new underground cavern seepage influencing factor data into the depth-physical-constrained geographic-weighted regression model to obtain the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of the total cavern seepage volume, the cavern seepage volume dynamic prediction method further includes:

[0037] Extract the geometric features of fractures and the seepage-related features from the seepage influencing factor data of the underground cavern;

[0038] Screening effective influencing factors of cavern seepage volume from the fracture geometry features and the seepage-related features;

[0039] Collinearity factors were eliminated using a variance inflation factor, and an influence factor matrix was constructed based on the remaining factors.

[0040] The deep physical constraint geographic weighted regression model is trained using the aforementioned influence factor matrix and measured infiltration volume to obtain a well-trained deep physical constraint geographic weighted regression model.

[0041] The new underground cavern seepage influencing factor data are input into the deep physical constraint geographic weighted regression model to obtain the predicted seepage volume for each fractal fracture element and the spatiotemporal distribution of the total cavern seepage volume, including:

[0042] The new data on seepage influencing factors in underground caverns are input into the trained deep physical constraint geographic weighted regression model to obtain the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of seepage volume in the entire cavern.

[0043] According to a second aspect of the present invention, a device for dynamically predicting cavern seepage is provided, comprising:

[0044] The data acquisition module is used to acquire data on the influencing factors of seepage in underground caverns;

[0045] The first modeling module is used to establish a dynamic evolution model of the fracture network based on the seepage influencing factor data of the underground cavern, so as to dynamically update the fractal characteristic parameters of each fractal fracture element.

[0046] The second modeling module is used to establish an anisotropic permeability tensor model based on the underground cavern seepage influencing factor data;

[0047] The third modeling module is used to construct a deep physical constraint geographic weighted regression model based on the anisotropic permeability tensor model and seepage core parameters.

[0048] The running module is used to input the new underground cavern seepage influencing factor data into the deep physical constraint geographic weighted regression model to obtain the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of the total cavern seepage volume.

[0049] The result output module is used to output the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of the total seepage volume of the cavern as the dynamic prediction result of the cavern seepage volume.

[0050] According to a third aspect of the present invention, an electronic device is provided, comprising: a processor; and a memory storing computer-readable instructions that, when executed by the processor, implement the method of the first aspect.

[0051] According to a fourth aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the method of the first aspect.

[0052] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0053] The dynamic prediction method for cavern seepage volume of the present invention acquires data on seepage influencing factors in underground caverns, incorporates key information such as hydrological recharge, fracture structure, and construction disturbance into the prediction input, and establishes a dynamic evolution model of fracture network based on the data to update the characteristics of fractal fracture elements in real time. At the same time, it constructs an anisotropic permeability tensor model to accurately characterize the differences in water conduction direction of fractures, and then establishes a depth-physical-constrained geographically weighted regression model in combination with the core seepage parameters. This model not only describes the spatial non-stationarity of seepage volume but also suppresses inverse physical fitting. Therefore, it can achieve rapid, stable, and physically consistent prediction output of the spatiotemporal distribution of seepage volume of each fractal fracture element and the entire cavern under continuously changing working conditions from excavation to operation. This provides quantitative support for early warning of water inrush risk, optimization of drainage design, and engineering safety management, while reducing the time and cost of traditional numerical iterative calculations.

[0054] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0055] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the specification, serve to explain the principles of the invention.

[0056] Figure 1 A schematic diagram of a system architecture for an exemplary application environment in which an embodiment of the present invention can be applied is shown;

[0057] Figure 2 The schematic diagram illustrates a process flow diagram of a method according to some embodiments of the present invention;

[0058] Figure 3 A schematic diagram of an apparatus according to some embodiments of the present invention is shown;

[0059] Figure 4 The schematic diagram illustrates the structure of a computer system of an electronic device according to some embodiments of the present invention.

[0060] Figure 5 A schematic diagram of a computer-readable storage medium according to some embodiments of the present invention is shown. Detailed Implementation

[0061] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention as detailed in the appended claims.

[0062] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The singular forms “a,” “the,” and “the” used in this invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0063] It should be understood that although the terms first, second, third, etc., may be used in this invention to describe various information, this information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, first information may also be referred to as second information without departing from the scope of this invention, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "when," "when," or "in response to a determination."

[0064] Figure 1 A schematic diagram of a system architecture for an exemplary application environment in which a method and apparatus for dynamic prediction of cavern seepage volume, according to embodiments of the present invention, can be applied.

[0065] like Figure 1 As shown, system architecture 100 may include one or more terminal devices such as desktop computer 101, portable computer 102, and smartphone 103, network 104, and server 105. Network 104 is used as a medium to provide a communication link between the terminal devices and server 105. Network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables. Terminal devices may be various electronic devices with data processing capabilities, which have a display screen for displaying the dynamic prediction results of cavern seepage to the user, including but not limited to the aforementioned desktop computer, portable computer, and smartphone. It should be understood that... Figure 1 The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, there can be any number of terminal devices, networks, and servers. For example, server 105 could be a sub-server cluster composed of multiple sub-servers.

[0066] The dynamic prediction method for cavern seepage provided in this embodiment of the invention can generally be executed by a terminal device, and correspondingly, the dynamic prediction device for cavern seepage is generally installed in the terminal device. However, it is readily understood by those skilled in the art that the dynamic prediction method for cavern seepage provided in this embodiment of the invention can also be executed by a server 105, and correspondingly, the dynamic prediction device for cavern seepage can also be installed in the server 105. This exemplary embodiment does not impose any special limitations on this.

[0067] Furthermore, it should be understood that the dynamic prediction method for cavern seepage volume in this embodiment of the invention can be configured as a software module. In some implementation scenarios, the dynamic prediction scheme for cavern seepage volume of the present invention can be deployed independently to enable dynamic prediction of seepage volume for different caverns. In other implementation scenarios, the dynamic prediction scheme for cavern seepage volume of the present invention can be deployed within other software as a functional module of that software, such as in underground cavern analysis software. The present invention does not impose any particular restrictions on the application of the dynamic prediction method for cavern seepage volume.

[0068] The embodiments of the present invention will now be described in detail.

[0069] like Figure 2 As shown, Figure 2 This is a flowchart illustrating a method for dynamically predicting cavern seepage volume according to an exemplary embodiment of the present invention, comprising:

[0070] S210: Obtain data on the influencing factors of seepage in underground caverns;

[0071] S220: Based on the data of seepage influencing factors in the underground cavern, establish a dynamic evolution model of the fracture network to dynamically update the fractal characteristic parameters of each fractal fracture element;

[0072] S230: Establish an anisotropic permeability tensor model based on the data of seepage influencing factors in the underground caverns;

[0073] S240: Construct a deep physical constraint-based geographically weighted regression model based on the anisotropic permeability tensor model and seepage core parameters;

[0074] S250: Input the new underground cavern seepage influencing factor data into the deep physical constraint geographic weighted regression model to obtain the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of the total cavern seepage volume;

[0075] S260: Output the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of seepage volume of the entire cavern as the dynamic prediction result of the cavern seepage volume.

[0076] By acquiring data on influencing factors of underground cavern seepage, the information relied upon for seepage prediction is expanded from a single seepage monitoring quantity to a multi-dimensional input including hydrological recharge, fracture structure, and construction conditions. This reduces the impact of missing information and input bias on the prediction results and improves the interpretability and effectiveness of the prediction.

[0077] Based on the data on seepage influencing factors in underground caverns, a dynamic evolution model of the fracture network is established, and the fractal characteristic parameters of each fractal fracture element are dynamically updated. This allows the spatial heterogeneity of the fracture network and its temporal changes during the excavation-operation period to be explicitly characterized, thereby overcoming the problem that static fracture parameters are difficult to adapt to construction disturbances and changes in operating conditions, and improving the stability and consistency of seepage prediction in dynamic scenarios.

[0078] Based on the data of seepage influencing factors in underground caverns, an anisotropic permeability tensor model is established to quantify the directional differences in the water conductivity of fractured rock masses. This allows the permeability parameters to simultaneously reflect spatial and directional differences, thereby overcoming the problem that homogeneous / single-valued equivalent permeability coefficients are difficult to describe the anisotropy of fractures, leading to prediction bias. This improves the spatial accuracy and engineering verifiability of seepage prediction.

[0079] Based on the anisotropic permeability tensor model and seepage core parameters, a deep physical constraint-based geographically weighted regression model is constructed. This model not only describes the spatial non-stationarity of seepage but also introduces seepage physical constraints to suppress inverse physical fitting, thereby improving the robustness, generalization ability, and physical consistency of predictions under conditions of noise, missing data, or insufficient samples.

[0080] By inputting the new underground cavern seepage influencing factor data into the deep physical constraint geographic weighted regression model, the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of the total cavern seepage volume are obtained. This allows the prediction results to be updated quickly with new working conditions, avoiding the high computational cost caused by repeated reconstruction and iterative solutions of traditional numerical models. At the same time, it enables the identification of local high-risk seepage zones and supports the optimization of global drainage and waterproofing design.

[0081] The predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of seepage volume in the entire cavern are output as dynamic prediction results of cavern seepage volume. The prediction results can be directly used for early warning of water inrush risk, configuration of drainage capacity, adjustment of construction organization and safety management during operation, thereby reducing the risk of water inrush disaster and reducing the engineering cost redundancy caused by conservative design.

[0082] In S210, data on the influencing factors of seepage in underground caverns are obtained.

[0083] In this embodiment, the seepage influencing factor data of underground caverns is used to characterize the control conditions, media structure characteristics, construction disturbance process and seepage response results of the cavern seepage field. It includes at least traditional geological and hydrological data, construction data and monitoring data. The seepage influencing factor data is organized in a unified spatiotemporal coordinate system so as to subsequently establish a fracture network dynamic evolution model, anisotropic permeability tensor model and a deep physical constraint geographic weighted regression model.

[0084] Traditional geological and hydrological data are used to characterize the properties of the surrounding rock medium, groundwater recharge and boundary conditions, and fracture occurrence. This data may include, but is not limited to: borehole exploration data, such as borehole columnar sections, lithological stratification, and rock quality indicators. Designation (RQD), rock mass integrity indices, fracture development levels, permeability segmentation information corresponding to borehole depth, etc.; fracture sketches: such as fracture sketches of tunnel faces / walls, joint strike and dip statistics, fracture density, fracture length distribution, and identification of penetrating fractures, etc.; geophysical data: such as the location of anomalies, water-bearing structure indications, and fracture zone ranges obtained by seismic wave methods, electrical methods, transient electromagnetic methods, and ground-penetrating radar, etc.; pressure water test data: such as Lugeon values ​​at different depths or in different segments, distribution of equivalent permeability coefficients, and segmented permeability evaluation results, etc.; groundwater level data: such as well water levels, pore pressure monitoring data, and changes in recharge head boundaries, etc.; rainfall data: such as rainfall intensity, cumulative rainfall, rainfall duration, and hysteresis characteristics related to groundwater level / permeability response, etc.

[0085] Construction data is used to characterize the disturbance process of the surrounding rock stress field and fracture channels during excavation and its evolution over time. This data may include, but is not limited to, excavation steps: such as excavation methods (drill and blast / mechanical excavation), cycle advance, excavation cross-sectional shape, excavation sequence, and sectional excavation strategies; support measures: such as anchor bolt / cable arrangement, shotcrete thickness and age, steel arch frame installation, secondary lining construction time, and cutoff wall installation; and construction time-series data: such as changes in the face advance position over time, key construction node timestamps (excavation completion, initial support completion, grouting completion, etc.), blasting parameters, or disturbance intensity indices. In some implementations, the construction data may also include grouting parameters (grout mix ratio, grouting pressure, grouting volume, grouting section) and drainage measures parameters to characterize the dynamic impact of sealing and drainage on seepage channels.

[0086] Monitoring data is used to characterize the response state and seepage output of the surrounding rock during construction and operation. It may include, but is not limited to: rock mass stress: such as data from surrounding rock stress gauges, anchor bolt stress, support structure stress, surrounding rock deformation / convergence, etc., which can be used to reflect the trend of stress adjustment and crack opening change; seepage volume: such as time-series monitoring values ​​of seepage volume at different measuring points in the cavern, flow data of drainage ditches / drainage systems, etc., which are used as dependent variables or verification indicators for model training.

[0087] In some implementations, the underground cavern seepage influencing factor data can be obtained from the aforementioned traditional geological and hydrological data, construction data, and monitoring data after preprocessing. Specifically, the preprocessing includes at least: identifying and removing outliers from the original data to eliminate outlier interference caused by sensor drift, false alarm codes, or input errors; completing missing data to ensure data integrity at each time step / spatial unit; normalizing or standardizing data with different dimensions and value ranges to avoid unreasonable dominance of a single high-dimensional feature in model training; and performing spatiotemporal registration on data from different sources, mapping data such as borehole data, geophysical exploration data, fracture sketches, water pressure tests, groundwater levels, rainfall, excavation steps, support measures, construction sequence, rock stress, and seepage volume to a unified spatial coordinate and a unified time scale, thereby forming an underground cavern seepage influencing factor dataset that can be used for subsequent fracture network dynamic evolution modeling, anisotropic permeability tensor construction, and training / prediction of deep physical constraint geographic weighted regression models.

[0088] In S220, a dynamic evolution model of the fracture network is established based on the data of the seepage influencing factors of the underground cavern, so as to dynamically update the fractal characteristic parameters of each fractal fracture element.

[0089] In some embodiments, the data on seepage influencing factors in underground caverns includes rock mass stress monitoring data, which can be obtained by surrounding rock stress gauges, anchor stress gauges, steel arch stress monitoring devices and / or support stress monitoring devices. The rock mass stress monitoring data includes stress magnitude, stress direction and / or time-series data on stress changes over time.

[0090] A dynamic evolution model of the fracture network is established based on the data of seepage influencing factors in the underground cavern, so as to dynamically update the fractal characteristic parameters of each fractal fracture element. Specifically, it can be implemented as follows:

[0091] First, fracture geometric features are extracted from the seepage influencing factor data of the underground cavern. These fracture geometric features can be jointly determined by borehole logging data, fracture sketches, and geophysical interpretation results. The fracture geometric features include at least one or more of the following: fracture orientation, fracture dip angle, fracture length, fracture spacing, fracture density, fracture aperture, and fracture connectivity. To facilitate subsequent modeling, the fracture geometric features can be spatially mapped according to the cavern's axial mileage, cross-sectional location, and surrounding rock depth, and then temporally aligned at a predetermined time step (e.g., by day, week, or construction cycle) to obtain a fracture feature sequence or fracture feature field for multi-scale analysis.

[0092] Secondly, the fractal characteristic parameters in the fracture geometry are extracted using the multifractal detrended fluctuation analysis (MF-DFA) method. In one example, the fracture density sequence, fracture spacing sequence, and / or fracture aperture sequence within a spatial window or a candidate cell are used as inputs to MF-DFA. Through detrending processing, scale segmentation, fluctuation function calculation, and scaling exponent fitting, the fractal dimension related to multi-scale heterogeneity is obtained. Multifractal spectral parameters (including generalized fractal dimension) And the Strange Spectrum ) and / or connectivity Iso-fractal characteristic parameters. These fractal characteristic parameters can characterize the complexity of the fracture network at different scales and the local enrichment / sparseness differences, providing a structural basis for subsequent fractal fracture element division and dynamic updating.

[0093] Traditional fracture fractal dimension The result was obtained using the box counting method. In one example, a box with a side length of [missing information] was selected. The grid covers the statistical domain, and the number of boxes required to cover all cracks is statistically significant. And calculate the fractal dimension of the fracture according to the following formula. :

[0094]

[0095] in, To calculate the side length of the box (unit: m), the value can be manually determined in engineering implementation, and the range is as follows: m, and can be adjusted according to the density of the cracks; The number of boxes required to cover all cracks can be calculated by statistical analysis of crack sketches or by specialized software. Used to reflect the complexity of fracture development. The larger the value, the more developed the fissures, the denser the seepage channels, and the stronger the equivalent permeability of the rock mass.

[0096] Fractal connectivity The ability of a fracture to form a continuous seepage channel is characterized by the following formula:

[0097]

[0098] in, The total length of the through fracture that can form a continuous seepage channel (unit: m). The total length (in meters) of all cracks within the statistical domain is given by the following. and It can be obtained through geological logging, fracture network diagrams, or fracture sketches; ,and The larger the value, the better the fissure connectivity, the stronger the ability to form effective seepage channels, and the stronger the seepage capacity.

[0099] In one example, the quality index function is obtained based on MF-DFA. And calculate:

[0100]

[0101] And by Obtain the Singularity Index With the Strange Spectrum:

[0102]

[0103]

[0104] in, The order parameter is used to adjust the contribution of fracture sets of different intensities to multi-scale statistics; and This is used to characterize the multi-scale distribution inhomogeneity and structural complexity of fracture networks. By... and The combination can simultaneously characterize the fracture network from three dimensions: multifractal heterogeneity, overall geometric complexity, and channel connectivity, thus providing feature support for subsequent fractal fracture element division, fracture dynamic update, and permeability assignment.

[0105] Then, based on the fractal feature parameters, a clustering algorithm is used to divide the underground cavern into several fractal fracture elements that are internally homogeneous and externally significantly different. In some implementations, each candidate spatial unit (e.g., a spatial unit formed by mileage segments along the cavern axis and grid units along the cross-section) is first taken as the object, and multifractal spectrum parameters and traditional fractal / connectivity indices are extracted from the fracture information corresponding to the candidate spatial unit to form a clustering feature set. For example, in one example, for the first... Construct feature vectors from candidate spatial units Each feature can be standardized to eliminate the influence of differences in dimensions and scales on the clustering results. Then, the underground cavern is segmented along the axial direction (e.g., segmented by equidistant distance or by construction cycle), and within each mileage segment, it is discretized into a grid along the cross-sectional direction to obtain a set of candidate spatial units covering the cavern space; and the feature vectors are used to... As input to the clustering algorithm, based on the principle of density attainability and density connectivity, candidate spatial units with similar fractal characteristics and relatively continuous spatial distribution are clustered into the same class, while discrete units that cannot be classified into any high-density class are identified as noise points or boundary points. Through the density clustering mechanism of the clustering algorithm, partitions with significant differences in fracture fractal characteristics can be automatically identified without pre-specifying the number of clusters, suitable for engineering scenarios where underground cavern fractures exhibit non-uniform distributions such as local enrichment zones and fracture zones. Subsequently, the spatial region corresponding to each clustering result is defined as a fractal fracture element, making the fractal fracture element the basic unit for seepage prediction, and satisfying the requirements of internal homogeneity and significant external differences. To ensure relatively uniform permeability within the fractal fracture element, the following partitioning criteria are set in some implementations: for any fractal fracture element, the fractal characteristic parameters of its internal candidate spatial units are statistically analyzed, and the corresponding coefficient of variation is calculated. The coefficient of variation satisfies When a clustering result does not meet the aforementioned coefficient of variation threshold, it can be further subdivided by adjusting the parameters of the clustering algorithm (e.g., neighborhood radius and minimum number of samples), or further divided into smaller fractal fracture elements along the axial / cross-sectional direction, until the coefficient of variation threshold is met. This subdivision ensures that the fractal characteristic parameters within each fractal fracture element fluctuate minimally, making the rock mass permeability approximately uniform within the element. This provides a reliable foundation for subsequent dynamic permeability coefficient assignment based on element differences and local modeling using geographic weighted regression.

[0106] Finally, based on the rock mass stress monitoring data and the effective stress principle, the change in rock mass stress during construction excavation and support is calculated, and the variation law of fracture aperture is derived to dynamically update the fractal characteristic parameters of each fractal fracture element. In one example, for the first... A fractal fracture element, at time... Obtain rock mass stress monitoring values ​​and calculate the stress change compared to the previous time step. Based on the effective stress principle, the total stress change is converted into the effective stress change. The effective stress can be obtained by subtracting the pore water pressure from the total stress; subsequently, the updated amount of the fracture aperture is determined based on the fracture normal stress-aperture response relationship (which can be a linear compressibility relationship, an exponential closure relationship, or an empirical fitting relationship). This results in an increase in crack aperture when the effective stress decreases due to excavation unloading or support stress adjustment, and a decrease in crack aperture when the effective stress increases due to support compaction or grouting consolidation. Furthermore, the updated crack aperture is used to correct the crack geometric feature sequence (e.g., aperture sequence, equivalent connectivity index), and the corresponding fractal feature parameters are recalculated or recursively updated, thereby achieving dynamic updating of the fractal feature parameters of the fractal crack element with construction disturbance and stress evolution.

[0107] Through the above implementation methods, the dynamic evolution model of the fracture network can link the multi-scale structural features of fractures with stress changes caused by construction disturbances, realize the continuous updating of fractal fracture element features, and enable the subsequent anisotropic permeability tensor model and deep physical constraint geographic weighted regression model to adopt time-varying fracture structure inputs that are more in line with engineering practice, thereby improving the accuracy and stability of seepage prediction during the construction and operation periods.

[0108] In S230, an anisotropic permeability tensor model is established based on the data of the seepage influencing factors of the underground cavern.

[0109] The data on seepage influencing factors in underground caverns includes fracture attitude data. This fracture attitude data characterizes the spatial orientation of fractures and includes at least one of fracture strike and dip angle, preferably simultaneously including fracture strike, dip angle, and dip direction. The fracture attitude data can be obtained from geophysical interpretation results such as borehole logging, cavern wall / face fracture sketches, 3D laser scanning, photogrammetric modeling, and / or ground-penetrating radar, and may further include fracture group classification results and statistical distribution parameters of each group (e.g., average strike, average dip angle, dispersion). In some embodiments, the fracture attitude data is summarized and statistically analyzed according to fractal fracture elements to obtain the dominant fracture orientation and its directional distribution for each fractal fracture element, used to determine the dominant seepage direction and provide a basis for determining the principal axis direction of the anisotropic permeability tensor model.

[0110] In some implementations, firstly, within each fractal fracture element, the dominant orientation parameter and permeability parameter of the fracture are determined based on the underground cavern seepage influence factor data. The dominant orientation parameter includes at least the dominant fracture orientation, dominant dip angle, and their directional statistical distribution. The permeability parameter includes at least fracture aperture, fracture density, fracture connectivity, and / or fractal dimension. Then, the dominant orientation parameter is used to determine the principal axis direction of anisotropic permeability, and the permeability parameter is used to determine the permeability components of anisotropic permeability along each principal axis direction, thereby constructing an anisotropic permeability tensor at the fractal fracture element scale.

[0111] In one example, for the first A fractal fracture element is used to construct a local coordinate system, such that the first principal axis is aligned with the dominant water-conducting direction of the fracture, the second principal axis is orthogonal to the first principal axis, and the third principal axis is along the radial or vertical direction of the cavern. Within this local coordinate system, the principal values ​​of the anisotropic permeability tensor are determined. , and The principal value can be determined based on the combined characteristics of the equivalent permeability coefficient, fracture aperture, and connectivity obtained from pressure water tests, as well as fractal characteristic parameters. The conversion from fractal characteristic parameters to the principal permeability value can be achieved using power function coupling or empirical mapping relationships. Subsequently, the principal permeability value and the principal axis direction are assembled into a permeability tensor in a global coordinate system through rotation transformation. In order to obtain the first Anisotropic permeability tensor model of a fractal fracture element.

[0112] Furthermore, at different time steps during the construction or operation period, the fracture aperture and connectivity can be updated by combining rock mass stress monitoring data, groundwater level changes, and construction disturbance intensity, and the principal value of the anisotropic permeability tensor can be updated simultaneously. This allows the anisotropic permeability tensor model to dynamically evolve with changes in operating conditions, thereby providing a time-varying, directionally relevant, and geologically compatible permeability parameter field for subsequent calculation of core seepage parameters and depth-physically constrained geographic weighted regression models.

[0113] In some implementations, the anisotropic permeability tensor model is represented in three-dimensional tensor form as follows:

[0114]

[0115] in, ( () represents the components of the permeability tensor, with units of m / d. Specifically, ( The permeability component in the corresponding principal direction is used to characterize the difference in water conductivity of the rock mass in different principal directions (e.g., the dominant seepage direction, the direction perpendicular to the dominant seepage direction, and the third orthogonal direction). Typically, the dominant seepage direction can be determined by statistical analysis of the dominant strike / dip angle of fractures within fractal fracture elements, or by seepage observations and geological structural indications.

[0116] also, ( The permeability coupling effect coefficient reflects the coupling effect and mutual influence between seepage processes in different directions, such as anisotropic coupling effects caused by the intersection of fracture systems, non-orthogonal water-conducting channels, or structural shear. In engineering implementation, the coupling effect coefficient can be determined by fitting field-measured permeability coefficient data (e.g., pressure water tests, pumping tests, or inverted seepage parameters), and can be corrected as construction / operation data are updated to make the permeability tensor model more closely fit actual seepage conditions.

[0117] In S240, a deep physical constraint geographic weighted regression model is constructed based on the anisotropic permeability tensor model and the core parameters of seepage.

[0118] In some implementations, the core seepage parameters include at least the hydraulic gradient. With the cross-sectional area of ​​water passage .

[0119] Among them, hydraulic gradient The head loss per unit seepage path length is characterized by the following formula:

[0120]

[0121] In the formula, The difference in water head at both ends of the fractal fracture element can be calculated based on the elevation benchmark determined by the geological profile and combined with groundwater level data. The seepage path length can be determined based on the relationship between the geological profile and the spatial location of the cavern, or based on the equivalent path length of the dominant seepage direction within the fractal fracture element. This can be achieved by introducing a hydraulic gradient. It can explicitly quantify the impact of groundwater recharge intensity, water level boundary changes, and seepage driving force on seepage volume.

[0122] Cross-sectional area of ​​water flow The equivalent flow rate used to characterize seepage at the cavern-surrounding rock interface is calculated using the following formula:

[0123]

[0124] The perimeter of the cavern is determined by its designed shape; the length of the fractal fracture element is its length along the cavern's axis, which can be obtained from the spatial division of the fractal fracture element. This is achieved by introducing the cross-sectional area of ​​the water passage. The influence of differences in cavern shape and segment length on water collection capacity and seepage volume can be incorporated into the calculation of core seepage parameters, thereby providing a theoretical seepage volume. The physical constraint terms provide computable inputs that are consistent with engineering geometry.

[0125] Based on this, a deep physical constraint-based geographically weighted regression model is constructed using the anisotropic permeability tensor model and core seepage parameters. This model can be implemented as follows:

[0126] First, a power-law coupling formula for the permeability coefficient is constructed based on fractal fracture elements and their fractal characteristic parameters. In some implementations, the fractal characteristic parameters of each fractal fracture element are used as independent variables to establish a power-law coupling relationship between the permeability coefficient and the fractal characteristics, which is used to map the multi-scale structural features of the fracture into equivalent permeability. For example, the formula can be applied to the fractal fracture element... The fractal fracture element in the principal direction coordinate system is the first... Permeability components in each main direction The following power-law coupling form is adopted:

[0127]

[0128] in, For the first The benchmark permeability coefficient for each main direction, The power-law coefficient can be calibrated using pressure tests, pumping tests, or historical inversion data; in some implementations, it can also be... The above coupling formula incorporates statistical measures such as width and peak value to enhance the characterization of multifractal heterogeneity of cracks.

[0129] Secondly, the anisotropic permeability tensor model is assigned values ​​according to the power coupling formula of the permeability coefficient to obtain the anisotropic permeability tensor of each fractal fracture element. In some embodiments, for the first... For each fractal fracture element, first determine the tensor diagonal element in the principal direction coordinate system. (corresponding to the dominant seepage direction, the direction perpendicular to the dominant direction, and the third orthogonal direction, respectively), and calculated using the aforementioned power coupling formula; simultaneously, off-diagonal elements As a permeability coupling effect coefficient, it reflects the mutual influence of seepage in different directions. The off-diagonal element can be determined by fitting field-measured permeability coefficient data. Thus, the first... Anisotropic permeability tensor of a fractal fracture element:

[0130] Its unit is m / d

[0131] Then, the theoretical seepage volume of each fractal fracture element is calculated using the anisotropic permeability tensor of each fractal fracture element and the seepage core parameters. In some embodiments, water cut is introduced. To reflect the characteristics of saturated-unsaturated seepage, among which This can be obtained by combining the van Genuchten unsaturated seepage model with pore pressure / water content monitoring or boundary conditions. Based on the extension of Darcy's law, the following can be obtained: Theoretical water permeation of a fractal fracture element :

[0132]

[0133] in, Indicates the first The equivalent permeability component of the permeability tensor of each fractal fracture element in the seepage direction (in engineering implementation, the principal value in the dominant seepage direction or the value proportional to the direction cosine can be taken) (obtained by equivalent projection) The unit is m³ / d.

[0134] Next, spatial weights are constructed and a spatial weight matrix is ​​formed. In some implementations, a Gaussian kernel function is used to calculate the basic spatial weights between different fractal fracture elements, and construction disturbance factors and geological similarity factors are introduced to dynamically adjust the basic spatial weights, thereby obtaining the spatial weights. :

[0135]

[0136] in, For the first The and the first Spatial Euclidean distance (unit: m) between the centers of each fractal fracture element. For bandwidth parameters; This is the construction disturbance weight, used to reflect the disturbance impact of excavation face advancement, support, and grouting on the water conductivity of the fractures. It is preferably dynamically determined based on the distance from the fractal fracture element to the excavation face, such that the closer the distance, the greater the impact. The larger; Geological similarity weights are used to reflect the stronger correlation between units with similar lithology, fractal characteristics, and fracture parameters. The optimal weights are based on lithology category and fractal dimension. Connectivity The similarity metric with the multifractal spectrum parameters is determined so that the higher the similarity, the better. The larger the value, the better. To ensure that the weights can be used for local regression solutions, in some implementations, the weights are... Perform normalization processing to make it consistent with the first The sum of the weights corresponding to each fractal fracture element is 1, that is:

[0137]

[0138] in Let be the total number of fractal fracture elements. Further, the normalized weights are constructed as a subset of the total number of elements. The fractal fracture element corresponds to 3D diagonal weight matrix:

[0139]

[0140] To achieve spatially localized weighted regression where near elements are larger than far elements, fractal fracture elements with closer spatial locations or more similar geological structures are paired with the first element. The local coefficients of each fractal fracture element contribute more significantly.

[0141] Finally, based on the theoretical seepage volume and spatial weight matrix of each fractal fracture element, a deep physical constraint-based geographically weighted regression model is established, using a geographically weighted regression model as the foundation. In some implementations, the first... The effective influence factors of each fractal fracture element are composed of a vector. and based on the actual measured seepage volume As the dependent variable, a deep physical-constrained geographic-weighted regression model is established:

[0142]

[0143] in, This is a spatial local constant term (taken as 0 in some implementations). For the first The fractal fracture element pairs with the first Spatial local regression coefficients of each influencing factor This is the residual term. The spatial local regression coefficients are calculated using weighted least squares and a weight matrix. The solution can be specifically expressed as:

[0144]

[0145] in, For the first The coefficient vector of each fractal fracture element. For an effective influence factor matrix, This is the measured seepage vector. As an adaptive physical constraint weight, used to dynamically balance the contribution of data-driven regression fitting and physical constraints (theoretical seepage), it is employed in some implementations as follows:

[0146]

[0147] The maximum weight value, For the weight growth rate, The number of training iterations for the model; and This can be determined through error feedback and convergence optimization during the training process. (This is achieved by determining the theoretical seepage volume.) Introducing local regression as a physical constraint term can enhance physical consistency while maintaining the spatial non-stationarity of GWR, thereby suppressing inverse physical fitting and improving the stability and engineering applicability of seepage prediction.

[0148] In S250, the new underground cavern seepage influencing factor data are input into the deep physical constraint geographic weighted regression model to obtain the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of the total cavern seepage volume.

[0149] New underground cavern seepage influencing factor data refers to updated data collected at a new time step or a new construction stage, relative to the data used during the model training phase. This updated data includes at least one or more of the following: groundwater level, rainfall, construction progress location and support / grouting sequence, and seepage volume or seepage-related monitoring quantities. Before inputting the updated data into the model, it is preferable to perform outlier removal, missing value completion, normalization, and spatiotemporal registration on the updated data according to the data preprocessing procedure in step S210. This ensures that the updated data can be mapped to a unified spatiotemporal framework of each fractal fracture element, forming the first... A fractal fracture element at time... Influence factor vector .

[0150] In some implementations, the predicted values ​​of each fractal fracture element can be projected and stitched together according to the axial mileage and cross-sectional spatial position of the cavern to obtain the seepage distribution curve and cross-sectional distribution map of the cavern along the axial direction; furthermore, the predicted values ​​of all fractal fracture elements at the same time can be summarized to obtain the predicted value of the total seepage of the entire cavern. This results in a spatiotemporal dynamic prediction of fractal fracture element-level prediction and cavern-level aggregation.

[0151] In some implementations, before inputting the new underground cavern seepage influencing factor data into the deep physical constraint geographic weighted regression model to obtain the predicted seepage volume for each fractal fracture element and the spatiotemporal distribution of the total cavern seepage volume, the cavern seepage volume dynamic prediction method further includes steps of screening influencing factors and training the model, as follows:

[0152] First, the geometric features of fractures and the seepage-related features are extracted from the data on seepage influencing factors in the underground cavern. The geometric features of fractures may include fracture orientation, dip angle, length, spacing, density, aperture, connectivity, and fractal dimension / multifractal spectrum parameters. The seepage-related features may include groundwater level, rainfall, pore water pressure, water content, equivalent permeability coefficient from a pressure test, construction disturbance intensity index, rock mass stress variation, hydraulic gradient, and cross-sectional area of ​​the water passage. To ensure that the features can be used for regression modeling, it is preferable to summarize and statistically register the above features according to fractal fracture elements and time steps, forming a candidate feature set with fractal fracture elements as the sample granularity.

[0153] Secondly, effective influencing factors on cavern seepage volume are screened from the aforementioned fracture geometry features and seepage-related features. In some embodiments, correlation analysis, significance testing, stepwise regression, or feature selection methods based on information criteria can be used to screen candidate features, eliminating features with weak correlation or poor stability with seepage volume, and retaining effective influencing factors that can explain changes in seepage volume. These effective influencing factors serve as input variables for geographically weighted regression, characterizing the intensity of the effects of seepage driving force, permeability, and construction disturbance on seepage volume at different spatial locations.

[0154] Then, the variance inflation factor (VIF) is used to eliminate collinear factors, and an impact factor matrix is ​​constructed based on the remaining factors. In some implementations, the VIF value is calculated for each of the selected candidate effective impact factors. When the VIF value of an impact factor exceeds a preset threshold (e.g., 10 or 5), it is identified as a collinear factor and removed from the candidate set or merged with related factors. After processing the collinear factors, the remaining impact factors are used to construct an impact factor matrix in a sample-factor manner. Each row corresponds to a sample of a fractal fracture element at a certain time step, and each column corresponds to an influence factor.

[0155] Furthermore, the deep physical constraint geographic weighted regression model is trained using the aforementioned influence factor matrix and measured infiltration volume to obtain a trained deep physical constraint geographic weighted regression model. In some embodiments, the measured infiltration volume is used as the dependent variable vector. By combining the spatial weight matrix and the physical constraint term of theoretical infiltration, a weighted least squares and adaptive physical constraint weight strategy is adopted to solve the parameters and iteratively optimize the deep physical constraint geographic weighted regression model. This yields a set of model parameters that converge on the training data and meet the error index requirements, thus obtaining a well-trained deep physical constraint geographic weighted regression model.

[0156] Finally, the new underground cavern seepage influencing factor data is input into the deep physical constraint geographic weighted regression model to obtain the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of the total cavern seepage volume. This includes generating an input vector / input matrix consistent with the influencing factor matrix by processing the new underground cavern seepage influencing factor data according to the same preprocessing, feature extraction, and factor construction process as in the training phase, and inputting the new underground cavern seepage influencing factor data into the trained deep physical constraint geographic weighted regression model, thereby outputting the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of the total cavern seepage volume obtained by summing the predicted values.

[0157] In S260, the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of the total seepage volume of the cavern are output as the dynamic prediction result of the cavern seepage volume.

[0158] In some implementations, the output is provided in the form of electronic data, which includes at least tabular output and spatially distributed output. Tabular output records the predicted value of each fractal fracture element at different time steps and its corresponding influencing factor status; spatially distributed output presents the spatiotemporal distribution of cavern seepage in the form of curves, heat maps, or zonal color maps. Through these output methods, the prediction results can be directly used for cavern drainage and waterproofing design optimization, inrush risk early warning, and engineering safety management decisions, realizing the engineering application of dynamic prediction results of cavern seepage.

[0159] This invention selects the underground main powerhouse cavern of a large pumped power station as the experimental site. The excavated dimensions of the underground main powerhouse cavern are 110.0m × 25.4m × 55.5m (length × width × height). The overlying rock mass at this location is approximately 313m to 464m thick, with exposed bedrock, mostly strongly weathered layers, and dense vegetation. Geological findings reveal uneven fissure development, with localized through-fissure zones. The seepage characteristics exhibit significant anisotropy and dynamic changes (significantly affected by construction excavation disturbance), consistent with typical characteristics of underground engineering in complex fissured rock masses. Groundwater is mainly seepage to dripping, with the estimated groundwater level in the powerhouse area at a depth of approximately 180m, and a water level difference of approximately 235m above the powerhouse roof. The rock in the area is relatively hard, with well-developed fissures, and the surrounding rock is relatively intact to intact, with some areas being quite fractured. The surrounding rock type is mainly Class III, with some areas of dense faults or fissures classified as Class IV. During the construction period, 15 seepage monitoring points and 5 rock mass stress monitoring points were set up; 180 days of continuous measured data were collected, covering all types of data such as geological and hydrological data, construction disturbance data, and seepage monitoring data, providing sufficient data support for verification.

[0160] The actual infiltration volume at a certain monitoring point over 30 days was compared with the predicted value from the model of this invention. Some data are shown in the table below:

[0161] Table 1. Comparison of Actual and Predicted Values ​​at a Certain Monitoring Point

[0162]

[0163] In addition, if we select existing monitoring values ​​at a certain location and use a common data fitting method (least squares method) to fit the existing raw data, we can obtain the fitting coefficients and form the fitting equation as follows:

[0164]

[0165] Based on the above fitting equation, the cavern seepage volume for the next 10 and 20 days is predicted, as shown in Table 2. Obviously, it is just data fitting without considering physical equations and geological conditions, so the data prediction error is large, and the prediction error increases with the number of days to be predicted.

[0166] Table 2. Comparison of cavern seepage volume obtained using different methods within a preset number of days.

[0167]

[0168] Therefore, this invention can achieve rapid and stable prediction of cavern seepage under continuously changing working conditions during the excavation and operation period, providing quantitative support for early warning of water inrush risk, optimization of drainage design and engineering safety management, and reducing the time and cost of traditional numerical iterative calculations.

[0169] According to a second aspect of the present invention, a dynamic prediction device 300 for cavern seepage is also provided, with reference to Figure 3 As shown, the dynamic prediction device 300 for cavern seepage includes:

[0170] Data acquisition module 310 is used to acquire data on seepage influencing factors in underground caverns;

[0171] The first modeling module 320 is used to establish a dynamic evolution model of the fracture network based on the seepage influencing factor data of the underground cavern, so as to dynamically update the fractal characteristic parameters of each fractal fracture element.

[0172] The second modeling module 330 is used to establish an anisotropic permeability tensor model based on the underground cavern seepage influencing factor data;

[0173] The third modeling module 340 is used to construct a deep physical constraint geographic weighted regression model based on the anisotropic permeability tensor model and seepage core parameters.

[0174] The running module 350 is used to input the new underground cavern seepage influencing factor data into the deep physical constraint geographic weighted regression model to obtain the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of the total cavern seepage volume.

[0175] The result output module 360 ​​is used to output the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of the total seepage volume of the cavern as the dynamic prediction result of the cavern seepage volume.

[0176] It should be noted that although several modules of the dynamic prediction device for cavern seepage have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of the present invention, the features and functions of two or more modules described above can be embodied in one module or unit. Conversely, the features and functions of one module described above can be further divided into multiple modules or sub-modules for embodiment.

[0177] Furthermore, in an exemplary embodiment of the present invention, an electronic device capable of implementing the above-described method for dynamically predicting cavern seepage is also provided.

[0178] Those skilled in the art will understand that various aspects of the present invention can be implemented as systems, methods, or program products. Therefore, various aspects of the present invention can be specifically implemented as entirely hardware embodiments, entirely software embodiments (including firmware, microcode, etc.), or embodiments combining hardware and software aspects, collectively referred to herein as “circuit,” “module,” or “system.”

[0179] The following reference Figure 4 To describe an electronic device 400 according to such an embodiment of the present invention. Figure 4 The electronic device 400 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of the present invention.

[0180] like Figure 4 As shown, the electronic device 400 is manifested in the form of a general-purpose computing device. The components of the electronic device 400 may include, but are not limited to: at least one processing unit 410, at least one storage unit 420, a bus 430 connecting different system components (including storage unit 420 and processing unit 410), and a display unit 440.

[0181] The storage unit stores program code that can be executed by the processing unit 410, causing the processing unit 410 to perform the steps described in the "Exemplary Method" section above, based on various exemplary embodiments of the present invention. For example, the processing unit 410 can perform actions such as... Figure 2S210: Obtain data on seepage influencing factors in underground caverns; S220: Establish a dynamic evolution model of the fracture network based on the data on seepage influencing factors in underground caverns to dynamically update the fractal characteristic parameters of each fractal fracture element; S230: Establish an anisotropic permeability tensor model based on the data on seepage influencing factors in underground caverns; S240: Construct a deep physical constraint-based geographically weighted regression model based on the anisotropic permeability tensor model and the core seepage parameters; S250: Input the new data on seepage influencing factors in underground caverns into the deep physical constraint-based geographically weighted regression model to obtain the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of seepage volume in the entire cavern; S260: Output the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of seepage volume in the entire cavern as the dynamic prediction result of the cavern seepage volume.

[0182] Storage unit 420 may include readable media in the form of volatile storage units, such as random access memory (RAM) 421 and / or cache memory 422, and may further include read-only memory (ROM) 423.

[0183] Storage unit 420 may also include a program / utility 424 having a set (at least one) of program modules 425, including but not limited to: an operating system, one or more application programs, other program modules, and program data, each or some combination of these examples may include an implementation of a network environment.

[0184] Bus 430 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0185] Electronic device 400 can also communicate with one or more external devices 470 (e.g., keyboard, pointing device, Bluetooth device, etc.), and with one or more devices that enable a user to interact with electronic device 400, and / or with any device that enables electronic device 400 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 450. Furthermore, electronic device 400 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 460. As shown, network adapter 460 communicates with other modules of electronic device 400 via bus 430. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 400, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0186] Through the description of the above embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions of the embodiments of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, portable hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, terminal device, or network device, etc.) to execute the methods according to the embodiments of the present invention.

[0187] In exemplary embodiments of the present invention, a computer-readable storage medium is also provided, on which a program product capable of implementing the methods described above is stored. In some possible embodiments, various aspects of the present invention may also be implemented as a program product comprising program code, which, when the program product is run on a terminal device, causes the terminal device to perform the steps of the various exemplary embodiments of the present invention described in the "Exemplary Methods" section above.

[0188] refer to Figure 5 As shown, a program product 500 for implementing the above-described dynamic prediction method for cavern seepage volume according to an embodiment of the present invention is described. It may employ a portable compact disc read-only memory (CD-ROM) and include program code, and may run on a terminal device, such as a personal computer. However, the program product of the present invention is not limited thereto. In the present invention, the readable storage medium may be any tangible medium containing or storing a program that may be used by or in conjunction with an instruction execution system, apparatus, or device.

[0189] The program product may employ any combination of one or more readable storage media. Readable storage media may be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0190] Program code for performing the operations of this invention can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0191] Furthermore, the above figures are merely illustrative of the processes included in the method according to exemplary embodiments of the present invention, and are not intended to be limiting. It is readily understood that the processes shown in the above figures do not indicate or limit the temporal order of these processes. Additionally, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0192] Through the description of the above embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions of the embodiments of the present invention can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, portable hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, touch terminal, or network device, etc.) to execute the methods according to the embodiments of the present invention.

[0193] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. The invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the claims.

[0194] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for dynamically predicting cavern seepage, characterized in that, include: Obtain data on seepage influencing factors in underground caverns; A dynamic evolution model of the fracture network is established based on the data of seepage influencing factors in the underground cavern, so as to dynamically update the fractal characteristic parameters of each fractal fracture element; An anisotropic permeability tensor model was established based on the data of seepage influencing factors in the underground caverns. A deep physical constraint-based geographically weighted regression model is constructed based on the anisotropic permeability tensor model and the core parameters of seepage. The new data on seepage influencing factors in underground caverns are input into the deep physical constraint geographic weighted regression model to obtain the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of seepage volume in the entire cavern. The predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of seepage volume in the entire cavern are output as the dynamic prediction result of the cavern seepage volume. The data on seepage influencing factors in underground caverns includes rock mass stress monitoring data; A dynamic evolution model of the fracture network is established based on the seepage influencing factor data of the underground cavern, to dynamically update the fractal characteristic parameters of each fractal fracture element, including: Extract fracture geometric features from the seepage influencing factor data of the underground cavern; The fractal characteristic parameters in the fracture geometry are extracted using the multifractal detrending fluctuation analysis method. Based on the fractal characteristic parameters, a clustering algorithm is used to divide the underground cavern into several fractal fracture elements that are homogeneous internally and significantly different externally. Based on the rock mass stress monitoring data and combined with the effective stress principle, the change in rock mass stress during construction excavation and support is calculated, and then the change law of fracture aperture is derived, and the fractal characteristic parameters of each fractal fracture element are dynamically updated.

2. The method for dynamically predicting cavern seepage volume according to claim 1, characterized in that, The data on seepage influencing factors in underground caverns includes fracture orientation data; Based on the data on seepage influencing factors in underground caverns, an anisotropic permeability tensor model is established, including: The dominant seepage direction of the underground cavern is determined based on the fracture orientation data; Establish a principal direction coordinate system for permeability based on the dominant seepage direction; Establish the anisotropic permeability tensor model within the principal directional coordinate system.

3. The method for dynamically predicting cavern seepage volume according to claim 1, characterized in that, Based on the aforementioned anisotropic permeability tensor model and seepage core parameters, a deep physical constraint-based geographically weighted regression model is constructed, including: Based on the fractal fracture element and the fractal characteristic parameters of each fractal fracture element, a power coupling formula for the permeability coefficient is constructed. The anisotropic permeability tensor model is assigned values ​​according to the power coupling formula of the permeability coefficient to obtain the anisotropic permeability tensor of each fractal fracture element. The theoretical seepage volume of each fractal fracture element is calculated using the anisotropic permeability tensor of each fractal fracture element and the seepage core parameters. The basic spatial weights between different fractal fracture elements are calculated using a Gaussian kernel function. At the same time, construction disturbance factors and geological similarity factors are introduced to dynamically adjust the basic spatial weights to obtain the spatial weights. Based on the spatial weights of all fractal fracture elements, a spatial weight matrix is ​​obtained; Based on the theoretical seepage volume of each fractal fracture element and the spatial weight matrix, a deep physical constraint-based geographically weighted regression model is established, using a geographically weighted regression model as the foundation.

4. The method for dynamically predicting cavern seepage volume according to claim 3, characterized in that, The spatial weight include: in, For the first The fractal fracture element and the first The spatial Euclidean distance between the centers of each fractal fracture element. For bandwidth parameters; The construction disturbance weight is dynamically adjusted based on the distance between the excavation face and the fractal fracture element. The smaller The larger; The geological similarity weight is determined based on the similarity measure of the lithology and / or fracture characteristics of fractal fracture elements; the higher the similarity, the greater the weight. The larger.

5. The method for dynamically predicting cavern seepage volume according to claim 3, characterized in that, Based on the theoretical seepage volume of each fractal fracture element and the spatial weight matrix, and using a geographically weighted regression model as a foundation, the deep physical constraint-based geographically weighted regression model is established, including: in, For the first Measured water seepage of each fractal fracture element; For constant terms; For the first The fractal fracture element corresponding to the first One effective influencing factor, The number of effective impact factors; For the first The fractal fracture element corresponding to the first The spatial local regression coefficients are obtained by using the same method as the first... The spatial weight matrix corresponding to each fractal fracture element The weighted least squares solution is obtained, and it satisfies: , For the first The regression coefficient vector of each fractal fracture element For an effective influence factor matrix, This is the measured seepage vector; These are adaptive physical constraint weights used to balance the contribution of data-driven regression fitting and physical constraints, and satisfy the following conditions: , The maximum weight value and , For the weight growth rate, The number of training iterations for the model; For the first The theoretical seepage volume of each fractal fracture element, and satisfying the following conditions: in, For the first The components or equivalent permeability parameters of the anisotropic permeability tensor of a fractal fracture element. For hydraulic gradient, The cross-sectional area of ​​the water passage. The water content of the fractal fracture element.

6. The method for dynamically predicting cavern seepage volume according to any one of claims 1-5, characterized in that, Before inputting the new underground cavern seepage influencing factor data into the deep physical constraint geographic weighted regression model to obtain the predicted seepage volume for each fractal fracture element and the spatiotemporal distribution of the total cavern seepage volume, the cavern seepage dynamic prediction method further includes: Extract the geometric features of fractures and the seepage-related features from the seepage influencing factor data of the underground cavern; Screening effective influencing factors of cavern seepage volume from the fracture geometry features and the seepage-related features; Collinearity factors were eliminated using a variance inflation factor, and an influence factor matrix was constructed based on the remaining factors. The deep physical constraint geographic weighted regression model is trained using the aforementioned influence factor matrix and measured infiltration volume to obtain a well-trained deep physical constraint geographic weighted regression model. The new data on seepage influencing factors in underground caverns are input into the deep physical constraint geographic weighted regression model to obtain the predicted seepage volume for each fractal fracture element and the spatiotemporal distribution of seepage volume in the entire cavern, including: The new data on seepage influencing factors in underground caverns are input into the trained deep physical constraint geographic weighted regression model to obtain the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of seepage volume in the entire cavern.

7. A dynamic prediction device for cavern seepage volume using the dynamic prediction method for cavern seepage volume according to any one of claims 1-6, characterized in that, include: The data acquisition module is used to acquire data on the influencing factors of seepage in underground caverns; The first modeling module is used to establish a dynamic evolution model of the fracture network based on the seepage influencing factor data of the underground cavern, so as to dynamically update the fractal characteristic parameters of each fractal fracture element. The second modeling module is used to establish an anisotropic permeability tensor model based on the underground cavern seepage influencing factor data; The third modeling module is used to construct a deep physical constraint geographic weighted regression model based on the anisotropic permeability tensor model and seepage core parameters. The running module is used to input the new underground cavern seepage influencing factor data into the deep physical constraint geographic weighted regression model to obtain the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of the total cavern seepage volume. The result output module is used to output the predicted seepage volume of each fractal fracture element and the spatiotemporal distribution of the total seepage volume of the cavern as the dynamic prediction result of the cavern seepage volume.

8. An electronic device, characterized in that, include: processor; as well as A memory storing computer-readable instructions that, when executed by the processor, implement the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, It stores a computer program thereon, which, when executed by a processor, implements the method as described in any one of claims 1 to 6.

Images