Intelligent grouting simulation method and system based on adaptive sparse representation

Abstract

The present invention belongs to the technical field of grouting simulation, and specifically discloses an intelligent grouting simulation method and system based on adaptive sparse representation. The method includes: obtaining real-time data from the grouting construction site and performing preprocessing; constructing a multi-scale geological model based on the geological conditions of the target area, wherein the multi-scale geological model is obtained by coupling the overall geological model and the local detail model; using finite element analysis to grid the overall geological model, and using an adaptive sparse representation method to grid the local detail model; using real-time data from the grouting construction site to perform a preliminary simulation of the grouting process, comparing the simulation results with the actual grouting data, and automatically adjusting the model parameters when the deviation between the two exceeds a set threshold to continuously optimize the grouting simulation results. The present invention can significantly shorten the simulation calculation time and meet the needs of real-time monitoring and rapid decision-making.

Description

Technical Field

[0001] The present invention relates to the technical field of grouting simulation, and in particular to an intelligent grouting simulation method and system based on adaptive sparse representation. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] Grouting technology is widely used in underground engineering. Grouting is injected into foundations, rock fissures, or soil pores to improve foundation bearing capacity, control groundwater flow, or reinforce weak strata to meet engineering requirements. The traditional grouting process involves the flow, penetration, solidification, and diffusion of the slurry. However, these behaviors are influenced by a variety of complex factors, including the diversity of geological structures, the rheological properties of the slurry, and the control precision of the construction technology. Therefore, accurately simulating and predicting grouting effects has become a key challenge in engineering.

[0004] In practical engineering, geological structures often exhibit heterogeneous and multi-scale characteristics. For example, the fracture network of rock and soil may present minute pores at small scales but manifest as larger fracture zones at larger scales. After grouting, pressure gradually fills these gaps. Its flow path and diffusion rate vary with factors such as geological characteristics, grout viscosity, and grouting pressure, making construction control more difficult.

[0005] Traditional numerical simulation methods, such as the finite element method (FEM) and the finite difference method (FDM), are widely used in engineering simulation. However, when dealing with complex, multi-scale geological structures, the complex multi-scale geological structures and the nonlinear diffusion behavior of slurries require a large amount of computing resources, the calculations are long, and are not conducive to real-time applications; the finite element and finite difference methods have limitations in simulating the details of multi-scale geological structures, especially when the structure contains a large number of tiny cracks and pores. These methods find it difficult to accurately capture the actual diffusion path of the slurry, resulting in insufficient simulation accuracy; traditional methods are difficult to flexibly adjust in the face of changing construction conditions (such as slurry flow, pressure, etc.), and usually require complex model adjustments, resulting in low simulation efficiency, and difficult to provide real-time feedback under dynamic working conditions, and lack of adaptability. Summary of the Invention

[0006] To solve the above problems, the present invention proposes an intelligent grouting simulation method and system based on adaptive sparse representation, which divides the simulation area according to the different scales of the fluid diffusion process, thereby realizing multi-level precise processing of porous media; at the same time, the adaptive sparse representation technology is introduced, so that the system can dynamically adjust the solution accuracy according to the changes in the fluid diffusion characteristics.

[0007] In some embodiments, the following technical solutions are adopted:

[0008] An intelligent grouting simulation method based on adaptive sparse representation, comprising:

[0009] Acquire real-time data from the grouting construction site and perform pre-processing;

[0010] Based on the geological conditions of the target area, a multi-scale geological model is constructed, wherein the multi-scale geological model is obtained by coupling the overall geological model and the local detailed model; the overall geological model is meshed using finite element analysis, and the local detailed model is meshed using an adaptive sparse representation method;

[0011] The grouting process is preliminarily simulated using real-time data from the grouting construction site. The simulation results are compared with the actual grouting data. When the deviation between the two exceeds the set threshold, the model parameters are automatically adjusted to continuously optimize the grouting simulation results.

[0012] As a further solution, in the overall geological model, the overall fault and fracture zone structure is used as the simulation target; in the local detail model, the cracks and pores are modeled in detail, and multi-scale feature decomposition is completed to accurately simulate the slurry diffusion path.

[0013] As a further solution, the adaptive sparse representation method is used to mesh the local detail model. The specific process is as follows:

[0014] Generate a sparse dictionary of geological features through dictionary learning algorithm, and initialize the sparse coefficient vector x0 according to the input data;

[0015] Using the online gradient descent method, the sparse coefficient vector x is adaptively adjusted according to the actual construction data at each time step. t ;

[0016] All sparse coefficient vectors constitute a sparse coefficient matrix. The area where the non-zero sparse coefficients are dispersed in the sparse coefficient matrix is ​​regarded as a low-sparse area, and the area where the non-zero sparse coefficients are concentrated is regarded as a high-sparse area. The grid density of the low-sparse area is set to be greater than the grid density of the high-sparse area.

[0017] As a further solution, the online gradient descent method is used to adaptively adjust the sparse coefficient vector x according to the actual grouting data at each time step. t , specifically:

[0018] Assume that the input data y t The sparse representation at time step t is: t ≈Dx t , where D is a fixed dictionary matrix, x t is the sparse coefficient;

[0019] Construct the objective function based on the reconstruction error at the current time step t:

[0020] ;

[0021] Calculate the gradient of the objective function: ;in is the gradient of the reconstruction error with respect to the sparse coefficients, sign(x t ) is the gradient of the sparsity constraint;

[0022] Update the sparse coefficient x online using gradient descent t : ; where λ is the regularization parameter, η t is the learning rate.

[0023] As a further solution, after the construction data of the grouting construction site is updated, the dictionary D is updated using the K-SVD algorithm. Based on the updated dictionary D, the dictionary D and the sparse coefficient matrix X are iteratively updated again using the sparse coding method, and the grid density is re-divided based on the sparse coefficient matrix.

[0024] As a further solution, the real-time data of the grouting construction site was used to conduct a preliminary simulation of the grouting process. The specific process is as follows;

[0025] Set the initial conditions and boundary conditions for grouting;

[0026] For the overall geological model, the flow equation is: , solve the flow behavior of the grouting fluid in the soil and obtain the pressure, flow velocity and permeability coefficient at different positions in the soil layer; where k(r) is the permeability coefficient of the soil layer, which depends on the properties of the soil layer; p(r) is the pressure field of the grouting fluid; q(r) is the source term of the grouting fluid, usually expressed as the injection rate of the grouting fluid;

[0027] For the local detail model, through the diffusion equation:

[0028] ;

[0029] Determine the concentration distribution of the slurry in the soil pores; where C(r,t) is the slurry concentration at position r and time t; D(r) is the diffusion coefficient of the soil pores; is the gradient operator;

[0030] The time step Δt is set, and the flow and diffusion processes within each time step are gradually solved. The calculation results of each step will be used as the initial conditions for the next step until the iterative process is completed. Among them, in the calculation of each time step, the pressure, flow velocity and permeability coefficient calculated by the overall geological model are passed as input to the local detailed model. At the same time, the calculation results of the local detailed model are fed back to the overall geological model to adjust the parameters of the overall geological model.

[0031] As a further solution, the simulation results are compared with the actual grouting data. When the deviation between the two exceeds a set threshold, the model parameters are automatically adjusted. The model parameters include but are not limited to: permeability coefficient, grouting pressure, grouting flow rate, soil porosity, diffusion coefficient, slurry fluidity and viscosity. The adjustment process for the permeability coefficient is specifically as follows:

[0032] Define the error function E(k) to represent the difference between actual data and simulation results;

[0033] ;

[0034] Among them, p actual (x i ) is the actual measured pressure value, p sim (x i ,k) is the pressure value obtained by simulation calculation, k is the permeability coefficient; x i is the i-th column vector of the sparse coefficient matrix X.

[0035] By minimizing the error function E(k), the optimal permeability coefficient k is obtained opt , so that the deviation between the simulation results and the actual data is minimized; based on the obtained optimal permeability coefficient, the permeability coefficient in the model is updated.

[0036] In other embodiments, the following technical solutions are adopted:

[0037] An intelligent grouting simulation system based on adaptive sparse representation, comprising:

[0038] Data acquisition module, used to obtain real-time data of the grouting construction site and perform preprocessing;

[0039] An adaptive meshing module is used to construct a multi-scale geological model based on the geological conditions of the target area. The multi-scale geological model is obtained by coupling the overall geological model and the local detail model; the overall geological model is meshed using finite element analysis, and the local detail model is meshed using an adaptive sparse representation method;

[0040] The grouting simulation module is used to perform a preliminary simulation of the grouting process using real-time data from the grouting construction site, and compare the simulation results with the actual grouting data. When the deviation between the two exceeds the set threshold, the model parameters are automatically adjusted to continuously optimize the grouting simulation results.

[0041] In other embodiments, the following technical solutions are adopted:

[0042] A terminal device comprises a processor and a memory, wherein the processor is used to implement instructions; the memory is used to store multiple instructions, wherein the instructions are suitable for being loaded by the processor and executing the above-mentioned intelligent grouting simulation method based on adaptive sparse representation.

[0043] In other embodiments, the following technical solutions are adopted:

[0044] A computer-readable storage medium stores a plurality of instructions, wherein the instructions are suitable for being loaded by a processor of a terminal device and executing the above-mentioned intelligent grouting simulation method based on adaptive sparse representation.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] (1) By adopting a multi-scale model, the present invention can simultaneously consider both macrostructure and microscopic features. This multi-level modeling approach can accurately capture subtle changes within the geological body, such as the distribution of cracks and pores and the flow path of the slurry. Compared with traditional single-scale models, the present invention significantly improves the prediction accuracy in simulating the slurry diffusion and solidification process, thereby effectively reducing the construction risks caused by inaccurate predictions.

[0047] (2) This invention uses sparse representation technology, enabling the system to process high-dimensional geological data with low computational complexity. By sparsifying geological features, the system can efficiently store and calculate key information, reducing unnecessary data redundancy and computational burden. In large-scale engineering projects, simulation calculation time can be significantly shortened, meeting the needs of real-time monitoring and rapid decision-making.

[0048] (3) The present invention has a powerful dynamic feedback capability, which can monitor key parameters in the grouting process in real time and automatically adjust model parameters according to actual conditions. This real-time optimization mechanism ensures the flexibility and adaptability of the grouting process, and can promptly respond to various uncertainties that arise during the construction process, such as sudden changes in geological conditions or changes in slurry properties, ensuring the reliability of the construction effect.

[0049] (4) The intelligent simulation method of the present invention is applicable to various types of geotechnical engineering projects, including tunnel excavation, foundation reinforcement, groundwater control, etc. Under different geological conditions and engineering requirements, the system can flexibly adjust the model and parameters to provide personalized simulation services. This wide applicability makes the present invention highly valuable for promotion in actual engineering projects.

[0050] Other features and advantages of additional aspects of the present invention will be given in part in the following description and in part will become obvious from the following description or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is a flow chart of an intelligent grouting simulation method based on adaptive sparse representation in an embodiment of the present invention. DETAILED DESCRIPTION

[0052] It should be noted that the following detailed description is illustrative and is intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used in the present invention have the same meaning as commonly understood by those skilled in the art to which the present application belongs.

[0053] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0054] Example 1

[0055] In one or more embodiments, an intelligent grouting simulation method based on adaptive sparse representation is disclosed, combined with Figure 1 , specifically including the following process:

[0056] S101: Acquire real-time data of the grouting construction site and perform preprocessing.

[0057] In this embodiment, multiple sensors are used to collect real-time data from the grouting construction site, including grouting pressure, grouting flow rate, grouting velocity, and soil permeability change data. Drilling and seismic methods are used to collect on-site geological data, including soil structure, porosity, permeability coefficient, crack characteristics (width, direction, and connectivity), soil particle size and distribution, and geotechnical parameters (such as compression modulus and tensile strength). Furthermore, it is necessary to obtain information on the grouting material's viscosity, fluidity, curing time, the relationship between viscosity and shear stress, and the permeability characteristics of the slurry.

[0058] Since field data may be affected by noise, the system first performs data denoising, using algorithms such as wavelet transform and Kalman filtering to eliminate random noise, and then normalizes and standardizes the data to ensure the compatibility and consistency of the model with multi-source data, completing data preprocessing.

[0059] S102: Based on the geological conditions of the target area, a multi-scale geological model is constructed. The multi-scale geological model is obtained by coupling the overall geological model and the local detail model.

[0060] In this embodiment, at a macroscopic scale, an overall geological model is constructed with overall faults, fracture zones and other structures as simulation targets; the overall geological model includes parameters such as the overall structure of the soil layer, the permeability characteristics of the soil layer, porosity, and permeability coefficient, and simulates the large-scale flow and pressure distribution during the slurry injection process.

[0061] At the microscale, fine cracks and pores are modeled in detail to construct a localized detailed model. The porosity and pore distribution of the soil are determined, and a three-dimensional mesh is used to represent the pore structure. Each mesh element (e.g., a cube or octahedron) represents a small space between soil particles, simulating the flow of slurry. Depending on the needs of the grouting simulation, the mesh can be refined or coarsened, with a finer mesh used for critical areas (such as cracks and areas of high porosity) and a coarser mesh for less important areas.

[0062] S103: Using finite element analysis to mesh the overall geological model, and using an adaptive sparse representation method to mesh the local detail model.

[0063] To improve the adaptability and computational efficiency of the model, this example uses finite element analysis to perform an overall macrostructure partitioning, and based on this, a sparse representation method is used to characterize the microstructure in detail. By combining finite element analysis with sparse representation, a coarse mesh is applied to the overall geological model, while adaptive meshing using sparse representation is applied to the local detail model. This allows for accurate modeling of information at different scales, thus completing multi-scale feature decomposition and accurately simulating the slurry diffusion path.

[0064] As a specific example, the adaptive sparse representation method is used to mesh the local detail model. The specific process is as follows:

[0065] S1031: Discretize the data of the local detail model into an initial grid point set, represented by the matrix Y∈R m ×n , where m is the dimension of the physical property vector (such as porosity, particle density, etc.) at each grid point, and n is the number of initial grid points.

[0066] S1032: Normalize each column of Y to ensure that different physical parameters have consistent dimensions; use filtering or noise reduction technology to remove noise in the data to reduce interference with subsequent sparse representation.

[0067] S1033: Generate a sparse dictionary of geological features through a dictionary learning algorithm, and initialize a sparse coefficient vector x0 based on the input data;

[0068] Based on the geological characteristics and slurry flow behavior, an appropriate sparse dictionary (which can be constructed using wavelet or Fourier transforms) is selected to describe geological data of varying scales and characteristics. Initially, a basic dictionary is constructed based on field data and previous geological research results. This dictionary encompasses various geological features that may affect slurry diffusion (such as fracture characteristics and porosity distribution).

[0069] For each column of data point y i ∈Y (representing the feature vector of a grid point), in a fixed dictionary D∈R m×K , K represents the number of sparse bases (usually K>m), solve the following optimization problem:

[0070] ;

[0071] Assign a weight ω to each sample i , weight ω i Dynamically adjust according to regional importance (such as crack concentration or porosity); through weighted dictionary learning, the feature extraction ability of key areas can be enhanced.

[0072] λ is a regularization parameter that controls the balance between sparsity and error. Simply put, λ determines whether the model prioritizes accuracy (reducing error) or simplicity (sparseness) when fitting real data. A larger λ tends to make the model sparser, reducing the number of bases used; a smaller λ prioritizes data fitting accuracy and allows for more bases to be used.

[0073] In this embodiment, the method for selecting λ is as follows: ① Divide the dataset (construction data collected at different time steps during the grouting process) into k subsets (typically 5 or 10 subsets) by time step. ② For each k training and validation process, select one subset as the validation set and the remaining subsets as the training set. ③ Train the model on the training set and evaluate its performance on the validation set, calculating the error or loss (the difference between the simulation and the actual data). ④ Experiment with different λ values ​​and select the λ that minimizes the validation error, thereby optimizing the sparse representation and parameter adjustment during the simulation process.

[0074] Use the orthogonal matching pursuit (OMP) or Lasso algorithm to solve and obtain the initialized sparse coefficient vector x0, and update the sparse coefficient matrix X.

[0075] After determining the sparse dictionary D and the sparse coefficient matrix X, real-time geological characteristic data (such as fracture distribution and permeability) are input into the sparse representation module. Sparse coding algorithms (such as LASSO or OMP) are used to process the input geological characteristic data. The correlation between the input data and each basis in the sparse dictionary is calculated, and the basis with the highest correlation is selected. The input data is then reconstructed using a linear combination of these selected basis to achieve sparse representation. After the sparse processing, key geological features, such as the direction of the dominant fractures and the distribution pattern of the main pores, are extracted. These features are then used in subsequent grouting simulations.

[0076] In sparse representation, the correlation between the input data and each basis in the sparse dictionary is calculated to determine which basis contributes most to the reconstruction of the input data. This correlation is usually calculated based on a similarity measure between vectors, as follows:

[0077] Let the input data vector be y∈R n , the dictionary is D=[d1,d2,…,d k ]∈R n×k , where d i is the i-th basis vector in the dictionary, usually normalized to a unit vector (i.e. ||d i ||2=1), the sparse coefficient is x∈R k Input data y and dictionary basis vector d i Correlation r i It is usually expressed as the inner product of the two:

[0078] ;

[0079] in, Represents y and d i The dot product reflects the directional consistency of the two, and the absolute value This is because both positive and negative correlations can indicate strong correlations. If the dictionary basis vectors are not normalized, they need to be normalized: .

[0080] r i The larger the value, the greater the dictionary basis vector d i The closer to the direction of y, the greater the contribution. i Sort and select several bases with the highest correlation to represent y.

[0081] S1034: Using the online gradient descent method, the sparse coefficient vector x is adaptively adjusted according to the actual grouting process data at each time step. t , to ensure that the simulation results are consistent with the actual situation on site, thereby optimizing the grouting simulation accuracy.

[0082] Specifically, during the grouting process, the system periodically collects and processes on-site grouting data. This data includes grouting pressure, flow rate, soil permeability, and grouting liquid diffusion depth. At each time step, the system uses this data to adjust the sparsity coefficient of the current simulation, thereby improving simulation accuracy.

[0083] In this embodiment, in order to adapt to the changes in different geological conditions and construction stages, the sparse coefficient vector x is adaptively adjusted according to the actual grouting data of each time step. t The specific steps are as follows:

[0084] (1) Define the sparse representation model. Assume that the input data y t The sparse representation at time step t is: t ≈Dx t , where D is a fixed dictionary matrix, x t is a sparse coefficient vector, and D and y t It is known at every moment.

[0085] (2) Construct the objective function based on the reconstruction error of the current time step t:

[0086] ;

[0087] (3) For x t , calculate the gradient of the objective function: ;in is the gradient of the reconstruction error with respect to the sparse coefficients, sign(x t ) is the gradient of the sparsity constraint, and λ is the regularization parameter;

[0088] (4) Update the sparse coefficient x online using the gradient descent method t : ; Among them, η t is the learning rate, which determines the step size of each update; x t (k) is the sparse coefficient of the kth update, x t (k+1) is the sparse coefficient updated for the k+1th time.

[0089] At this point, the sparsity coefficient is dynamically updated at each time step based on the new geological conditions and construction environment. This allows the model's representation of these physical properties to be adjusted dynamically, making the simulation more realistic. At each time step, the sparsity coefficient is adjusted based on real-time construction data to more accurately reflect various parameters during the grouting process (such as grouting pressure and flow rate). The sparsity coefficient is a state parameter in the model at each moment, and is adjusted in real time at each time step using an online gradient descent method to better match construction data and ensure the real-time and accuracy of the simulation.

[0090] A key goal of sparse representation is to retain as few non-zero coefficients as possible through the sparse coefficient x. Therefore, adjusting the sparse coefficient usually results in a reduction in the number of features, maintaining the minimum features that contribute significantly to the target output (such as grouting simulation results). With the dynamic adjustment of the sparse coefficient (through online learning, i.e., the online gradient descent method mentioned above), the model will focus more on important features in the geological model and remove redundant features that have little impact on the simulation results. For example, under certain geological conditions, certain features (such as porosity, fracture density, etc.) may become more important, while other features (such as permeability) may become less important, so these unimportant features will be automatically eliminated through sparse representation.

[0091] If adjusting the sparsity factor causes a change in the distribution of features or important regions in the model, mesh optimization will be used to account for these changes during subsequent simulations. For example, after adjusting the sparsity factor, the model may focus more on high-density regions (such as areas with dense fractures or high porosity), which may result in the use of a finer mesh in these areas to capture more subtle physical variations.

[0092] Therefore, in this embodiment, all sparse coefficient vectors are formed into a sparse coefficient matrix, and the areas in the sparse coefficient matrix where non-zero sparse coefficients are dispersed (indicating large feature changes and complex structures) are taken as low-sparse areas, and the areas where non-zero sparse coefficients are concentrated (indicating relatively simple features and uniform structures) are taken as high-sparse areas; the grid density of the low-sparse areas is set to be greater than the grid density of the high-sparse areas; thereby realizing the grid division of the local detail model using the adaptive sparse representation method.

[0093] S104: Conduct a preliminary simulation of the grouting process using real-time data from the grouting construction site.

[0094] In this example, after meshing and discretization, initial conditions (i.e., initial pressure distribution and slurry concentration) are set based on geological data and the initial state of the grouting fluid. Boundary conditions (e.g., boundary pressure, slurry injection rate, porosity, and flow restrictions outside the soil layer) are also set based on the actual construction environment. A time step, Δt, is set, and the flow and diffusion processes within each time step are gradually solved. The results of each step serve as the initial conditions for the next step.

[0095] During the discretization process, the resulting solution system is usually a system of linear equations in the following form: AX=B, where A is the coefficient matrix containing information such as the flow equation and the diffusion equation; X is the vector of unknown variables (such as pressure and concentration); and B is the constant term, which is usually given by initial conditions, boundary conditions, etc.

[0096] The calculation results of the global geological model and the local detailed model are coupled to complete the final grouting behavior simulation. In each round of calculation, the pressure field, flow field, and permeability coefficient calculated by the global geological model are passed as input to the local detailed model. The calculation results of the local detailed model will affect the parameters of the global geological model. For example, if the local detailed model finds that the permeability of certain areas has changed (such as the grouting fluid diffuses to certain areas, causing the porosity of the soil to increase or decrease), the permeability coefficient or pressure field of the corresponding area needs to be adjusted in the global geological model. In multi-scale coupling, the coupling of the global geological model and the local detailed model is usually an iterative process. Each iteration transfers information between the global geological model and the local detailed model, and the model parameters are adjusted based on the feedback.

[0097] The following is a detailed calculation and iterative convergence judgment process:

[0098] In the overall geological model, the finite element method (FEM) is used to solve the flow behavior of the grouting fluid in the soil. The equation is: , where k(r) is the permeability of the soil layer, which depends on its properties; p(r) is the pressure field of the grouting fluid; and q(r) is the source term of the grouting fluid, typically expressed as the injection rate. By solving these equations, the overall geological model obtains information such as pressure distribution and flow velocity at different locations in the soil layer.

[0099] In the local detail model, the following diffusion equation is solved:

[0100] ;

[0101] Where C(r,t) is the slurry concentration at position r and time t; D(r) is the diffusion coefficient of the soil pores; is the gradient operator.

[0102] The local detail model calculates the concentration distribution of slurry in the soil pores by solving the diffusion equation.

[0103] Based on the adjustment results of the overall geological model, the overall geological model is solved again, and the new macroscopic results are used as input to the local detailed model to continue calculating the local detailed model. The results of the local detailed model are then fed back into the overall geological model again, and this process is repeated.

[0104] In each iteration, the parameter changes between the macro and local detailed models need to be checked to determine convergence. Common convergence criteria include: change amount check (at the end of each iteration, the change in the parameters of the global geological model and the local detailed model, such as the pressure field, concentration distribution, and permeability coefficient, is calculated. If the change is less than a preset threshold, convergence is considered) and error threshold (setting an error threshold. When the error of the model results falls below this threshold, the iteration is stopped).

[0105] In this embodiment, the convergence judgment standard can be set as: |X k+1 -X k | / |X k |<ε, where X k represents the result of the kth iteration, and ε is the preset convergence error threshold. When the difference between the results of two iterations is less than the preset error threshold, the iteration process is considered to have converged and the iteration can be stopped.

[0106] This embodiment can simulate and solve the grouting process more accurately through the coupled iterative solution of the overall geological model and the local detail model.

[0107] As an optional example, when the construction data of the grouting construction site is updated, the dictionary D is updated using the K-SVD algorithm. Based on the updated dictionary D, the dictionary D and the sparse coefficient matrix X are iteratively updated again using the sparse coding method, and the grid density is re-divided based on the sparse coefficient matrix.

[0108] Each basis in the dictionary represents a possible pattern, which can be different physical properties of the soil (such as pore structure and permeability) or different flow patterns during the grouting process. When real-time construction data is updated, the dictionary needs to be optimized to ensure it accurately represents the various changes in the grouting process. Dictionary updating is a global operation that affects the structure of the entire simulation. This embodiment uses the K-SVD algorithm to optimize the dictionary based on real-time data, thereby updating the representation capability of each basis. The dictionary updating process uses an iterative optimization method to ensure that the dictionary D better reflects the in-situ soil structure and grouting behavior. The updated dictionary D redirects the sparse coding process and updates the sparse coefficient matrix X, thereby obtaining more accurate simulation results. After the dictionary and sparse coefficient matrix are updated, the system re-meshes the grid based on these updated parameters. The change in grid density is based on the impact of the new model parameters on the soil microstructure and the grouting process, optimizing the mesh refinement and achieving more accurate simulation results. By updating the grid density, details of the grouting process can be better captured, especially in areas with significant changes in local grouting pressure or diffusion paths.

[0109] S105: Compare the simulation results with the actual grouting data. When the deviation between the two exceeds a set threshold, automatically adjust the model parameters to continuously optimize the grouting simulation results.

[0110] In this embodiment, during the simulation process, the deviation between the actual grouting process data (such as the actual flow pressure and flow rate of the slurry, and changes in soil porosity) and the simulation results is monitored in real time. If the deviation exceeds a set threshold, the system dynamically adjusts the model parameters using an adaptive algorithm to optimize the grouting speed and diffusion path, ensuring that the grouting process meets design requirements and construction standards. For example, if the slurry diffuses unevenly, the system will optimize the diffusion effect by increasing the pressure or adjusting the grouting flow rate.

[0111] The following are the adjustment processes of several main parameters:

[0112] (1) Adjustment of the permeability coefficient. By monitoring the diffusion range and pressure field of the grouting fluid in real time and comparing the actual permeability rate with the simulation results, the system can automatically adjust the permeability coefficient through optimization algorithms (such as the least squares method) so that the model can better reflect the actual permeability of the soil layer. First, define the error function E(k) to represent the difference between the actual data and the simulation results:

[0113] ;

[0114] Among them, p actual (x i ) is the actual measured pressure value, p sim (x i,k) is the pressure value obtained by simulation calculation, and k is the permeability coefficient. By minimizing the error function E(k), the optimal permeability coefficient k is obtained. opt , so that the deviation between the simulation results and the actual data is minimized: .

[0115] x i It is the i-th column vector of the sparse coefficient matrix X, which represents the sparse representation of the i-th data point (such as the measurement data at a certain moment or position) in the dictionary D basis.

[0116] Based on the calculated optimal permeability coefficient, the permeability coefficient in the model is updated, and the macro and local detail models are rerun.

[0117] (2) Adjustment of grouting pressure and flow rate. By measuring the flow rate, pressure, and velocity field of the grouting fluid in real time and comparing them with the simulation results, the difference between the actual flow rate and pressure and the simulated values ​​can be calculated. If the actual flow rate is low, it is possible to consider increasing the flow rate of the grouting fluid or increasing the grouting pressure.

[0118] (3) Adjustment of soil porosity. The system can automatically update the porosity parameters based on the settlement or porosity changes of the soil layer during the actual grouting process. By comparing the deviation between the actual porosity and the model prediction value, the porosity parameters in the overall geological model are adjusted.

[0119] (4) Adjustment of the diffusion coefficient. The diffusion coefficient in the local detail model is adjusted according to the real-time monitored slurry diffusion rate (through pressure changes in the soil, slurry concentration distribution, etc.).

[0120] The real-time optimization mechanism of this embodiment ensures the flexibility and adaptability of the grouting process, and can promptly respond to various uncertainties that arise during the construction process, such as sudden changes in geological conditions or changes in slurry properties, thereby ensuring the reliability of the construction effect.

[0121] Finally, the simulated results, such as grouting diffusion range, pressure distribution, and flow rate changes, are graphically displayed in 3D or dynamic graphs, helping users understand the grouting status in real time. If the grouting diffusion trend or pressure distribution exceeds the expected range, the system triggers a risk warning and recommends adjusting parameters (such as grouting flow rate and pressure) to reduce the risk. Real-time decision support is also provided, helping construction personnel to promptly assess grouting effectiveness and adjust grouting strategies to achieve optimal results.

[0122] Example 2

[0123] In one or more embodiments, an intelligent grouting simulation system based on adaptive sparse representation is disclosed, comprising:

[0124] Data acquisition module, used to obtain real-time data of the grouting construction site and perform preprocessing;

[0125] An adaptive meshing module is used to construct a multi-scale geological model based on the geological conditions of the target area. The multi-scale geological model is obtained by coupling the overall geological model and the local detail model; the overall geological model is meshed using finite element analysis, and the local detail model is meshed using an adaptive sparse representation method;

[0126] The grouting simulation module is used to perform a preliminary simulation of the grouting process using real-time data from the grouting construction site, and compare the simulation results with the actual grouting data. When the deviation between the two exceeds the set threshold, the model parameters are automatically adjusted to continuously optimize the grouting simulation results.

[0127] It should be noted that the specific implementation of the above modules is the same as that in Example 1 and will not be described in detail.

[0128] Example 3

[0129] In one or more embodiments, a terminal device is disclosed, which includes a processor and a memory, wherein the processor is used to implement instructions; the memory is used to store multiple instructions, and the instructions are suitable for the processor to load and execute the intelligent grouting simulation method based on adaptive sparse representation described in Example 1.

[0130] It should be understood that in this embodiment, the processor may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), off-the-shelf field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0131] The memory may include a read-only memory and a random access memory, and provides instructions and data to the processor. A portion of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0132] During implementation, each step of the above method may be completed by an integrated logic circuit of hardware in a processor or by instructions in the form of software.

[0133] Example 4

[0134] In one or more embodiments, a computer-readable storage medium is disclosed, in which a plurality of instructions are stored, wherein the instructions are suitable for being loaded by a processor of a terminal device and executing the intelligent grouting simulation method based on adaptive sparse representation described in Example 1.

[0135] Although the above describes the specific embodiments of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without any creative work are still within the scope of protection of the present invention.

Claims

1. An intelligent grouting simulation method based on adaptive sparse representation, characterized in that: include: Acquire real-time data from the grouting site and perform pre-processing; Based on the geological conditions of the target area, a multi-scale geological model is constructed, wherein the multi-scale geological model is obtained by coupling the overall geological model and the local detailed model; the overall geological model is meshed using finite element analysis, and the local detailed model is meshed using an adaptive sparse representation method; Use real-time data from the grouting site to conduct a preliminary simulation of the grouting process, compare the simulation results with the actual grouting data, and automatically adjust the model parameters when the deviation between the two exceeds the set threshold to continuously optimize the grouting simulation results; Among them, the adaptive sparse representation method is used to mesh the local detail model. The specific process is as follows: Generate a sparse dictionary of geological features through dictionary learning algorithm, and initialize the sparse coefficient vector x0 according to the input data; The sparse coefficient vector x is adaptively adjusted according to the actual grouting data at each time step using the online gradient descent method. t ; Specifically: Assume that the input data y t The sparse representation at time step t is: t ≈Dx t , where D is a fixed dictionary matrix, x t is the sparse coefficient; Construct the objective function based on the reconstruction error at the current time step t: ； Calculate the gradient of the objective function: ;in is the gradient of the reconstruction error with respect to the sparse coefficients, sign(x t ) is the gradient of the sparsity constraint; Update the sparse coefficient x online using gradient descent t : ; where λ is the regularization parameter, η t is the learning rate; All sparse coefficient vectors constitute a sparse coefficient matrix. The area where the non-zero sparse coefficients are dispersed in the sparse coefficient matrix is ​​regarded as a low-sparse area, and the area where the non-zero sparse coefficients are concentrated is regarded as a high-sparse area. The grid density of the low-sparse area is set to be greater than the grid density of the high-sparse area.

2. The intelligent grouting simulation method based on adaptive sparse representation according to claim 1, characterized in that: In the overall geological model, the overall fault and fracture zone structure is simulated; in the local detail model, the cracks and pores are modeled in detail, and multi-scale feature decomposition is completed to accurately simulate the slurry diffusion path.

3. The intelligent grouting simulation method based on adaptive sparse representation according to claim 1, characterized in that: After the grouting data at the grouting site is updated, the dictionary D is updated using the K-SVD algorithm. Based on the updated dictionary D, the dictionary D and the sparse coefficient matrix X are iteratively updated again using the sparse coding method, and the grid density is re-divided based on the sparse coefficient matrix.

4. The intelligent grouting simulation method based on adaptive sparse representation according to claim 1, characterized in that: The grouting process is preliminarily simulated using real-time data from the grouting site. The specific process is as follows: Set the initial conditions and boundary conditions for grouting; For the overall geological model, the flow equation is: , solve the flow behavior of the grouting fluid in the soil and obtain the pressure, flow velocity and permeability coefficient at different positions in the soil layer; where k(r) is the permeability coefficient of the soil layer, which depends on the properties of the soil layer; p(r) is the pressure field of the grouting fluid; q(r) is the source term of the grouting fluid, usually expressed as the injection rate of the grouting fluid; For the local detail model, through the diffusion equation: ； Determine the concentration distribution of the slurry in the soil pores; where C(r,t) is the slurry concentration at position r and time t; D(r) is the diffusion coefficient of the soil pores; is the gradient operator; The time step Δt is set, and the flow and diffusion processes within each time step are gradually solved. The calculation results of each step will be used as the initial conditions for the next step until the iterative process is completed. Among them, in the calculation of each time step, the pressure, flow velocity and permeability coefficient calculated by the overall geological model are passed as input to the local detailed model. At the same time, the calculation results of the local detailed model are fed back to the overall geological model to adjust the parameters of the overall geological model.

5. The intelligent grouting simulation method based on adaptive sparse representation according to claim 1, characterized in that: The simulation results are compared with the actual grouting data. When the deviation between the two exceeds the set threshold, the model parameters are automatically adjusted; the model parameters include but are not limited to: permeability coefficient, grouting pressure, grouting flow rate, soil porosity, diffusion coefficient, slurry fluidity and viscosity; The specific adjustment process of the permeability coefficient is as follows: Define the error function E(k) to represent the difference between actual data and simulation results; ； Among them, p actual (x i ) is the actual measured pressure value, p sim (x i ,k) is the pressure value obtained by simulation calculation, k is the permeability coefficient; x i is the i-th column vector of the sparse coefficient matrix X; By minimizing the error function E(k), the optimal permeability coefficient k is obtained opt , so that the deviation between the simulation results and the actual data is minimized; based on the obtained optimal permeability coefficient, the permeability coefficient in the model is updated.

6. An intelligent grouting simulation system based on adaptive sparse representation, characterized in that: include: Data acquisition module, used to obtain real-time data from the grouting site and perform preprocessing; An adaptive meshing module is used to construct a multi-scale geological model based on the geological conditions of the target area. The multi-scale geological model is obtained by coupling the overall geological model and the local detail model; the overall geological model is meshed using finite element analysis, and the local detail model is meshed using an adaptive sparse representation method; The grouting simulation module is used to perform a preliminary simulation of the grouting process using real-time data from the grouting site, and compare the simulation results with the actual grouting data. When the deviation between the two exceeds a set threshold, the model parameters are automatically adjusted to continuously optimize the grouting simulation results; Among them, the adaptive sparse representation method is used to mesh the local detail model. The specific process is as follows: Generate a sparse dictionary of geological features through dictionary learning algorithm, and initialize the sparse coefficient vector x0 according to the input data; The sparse coefficient vector x is adaptively adjusted according to the actual grouting data at each time step using the online gradient descent method. t ; Specifically: Assume that the input data y t The sparse representation at time step t is: t ≈Dx t , where D is a fixed dictionary matrix, x t is the sparse coefficient; Construct the objective function based on the reconstruction error at the current time step t: ； Calculate the gradient of the objective function: ;in is the gradient of the reconstruction error with respect to the sparse coefficients, sign(x t ) is the gradient of the sparsity constraint; Update the sparse coefficient x online using gradient descent t : ; where λ is the regularization parameter, η t is the learning rate; All sparse coefficient vectors constitute a sparse coefficient matrix. The area where the non-zero sparse coefficients are dispersed in the sparse coefficient matrix is ​​regarded as a low-sparse area, and the area where the non-zero sparse coefficients are concentrated is regarded as a high-sparse area. The grid density of the low-sparse area is set to be greater than the grid density of the high-sparse area.

7. A terminal device comprising a processor and a memory, wherein the processor is used to implement instructions; the memory is used to store multiple instructions, characterized in that: The instructions are suitable for being loaded by a processor and executing the intelligent grouting simulation method based on adaptive sparse representation according to any one of claims 1 to 5.

8. A computer-readable storage medium storing a plurality of instructions, characterized in that: The instructions are suitable for being loaded by a processor of a terminal device and executing the intelligent grouting simulation method based on adaptive sparse representation according to any one of claims 1 to 5.

Images