Numerical simulation method for building foundation grouting reinforcement repair

Abstract

The application relates to the technical field of building engineering, and discloses a numerical simulation method for building foundation grouting reinforcement and repair. The method divides the building foundation into three levels of material layer, structure layer and fluid layer, and establishes a numerical atlas model containing analysis nodes, associated edges and a dynamic attribute set. By acquiring real-time state parameters of the material layer, a material state characteristic vector is calculated, and based on this, the potential grouting influence domain of the fluid layer is dynamically positioned in a multilayer progressive manner. Multilayer cascade numerical simulation is performed in the influence domain, and an initial strategy set containing grouting pressure, material proportioning and path planning is generated. The system receives external input signals for dynamic correction, and updates the numerical atlas model according to feedback data after the strategy is executed. The method realizes dynamic deduction and strategy optimization of the grouting repair process, and enhances the accuracy and adaptability of foundation reinforcement.

Description

Technical Field

[0001] This invention relates to the field of building engineering technology, specifically to a numerical simulation method for grouting reinforcement and repair of building foundations. Background Technology

[0002] Over long-term use, building foundations are prone to settlement, cracking, or strength reduction due to load changes, groundwater erosion, and material aging, seriously affecting the safety and durability of the superstructure. Grouting reinforcement, a common foundation repair method, restores the integrity and load-bearing capacity of the foundation by injecting grout material into the foundation to fill voids and cement the soil. However, the grouting process is affected by multiple factors, including geological conditions, material properties, structural stress state, and fluid permeability. Traditional methods mainly rely on engineering experience and simplified models for grouting scheme design, which has significant limitations.

[0003] Currently used grouting simulation methods are mostly based on a single physical field or static assumptions, failing to fully reflect material nonlinearity, structure-fluid interaction, and dynamic changes during construction. For example, while the finite element method can simulate stress distribution, its characterization of grout flow in porous media is relatively coarse; and computational fluid dynamics methods focus on flow behavior but often ignore the time-varying response of soil structure and material properties. Furthermore, existing methods generally lack the ability to dynamically adjust grouting strategies, making online optimization based on real-time monitoring data or external commands difficult, leading to deviations between actual grouting results and design objectives.

[0004] Although some studies have attempted to incorporate multidisciplinary model coupling analysis, most methods remain in the offline computation stage, exhibiting low computational efficiency and weak integration, failing to effectively support rapid decision-making during construction. Especially under complex foundation conditions, the setting of grouting parameters such as pressure, mix proportions, and path still largely relies on human experience, lacking systematic and quantitative analysis tools. Therefore, there is an urgent need to develop a numerical simulation method capable of integrating multi-level information, possessing dynamic response and autonomous optimization capabilities, to improve the accuracy and reliability of grouting reinforcement. Summary of the Invention

[0005] The purpose of this invention is to provide a numerical simulation method for grouting reinforcement and repair of building foundations, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides a numerical simulation method for grouting reinforcement and repair of building foundations, the method comprising:

[0007] The building foundation to be repaired is deconstructed into three levels; wherein, the levels include a material layer, a structural layer, and a fluid layer;

[0008] A numerical graph model is established, which includes multiple analysis nodes, associated edges connecting the analysis nodes, and a dynamic attribute set for each analysis node. The dynamic attribute set includes material parameters, structural stress state, and fluid permeability coefficient.

[0009] Obtain the real-time state parameters of the material layer;

[0010] Based on the real-time state parameters of the material layer, calculate the material state feature vector of the material layer;

[0011] Based on the material state feature vector, the potential grouting influence domain of the fluid layer is dynamically located in the numerical spectrum model using a multi-layer progressive calculation method.

[0012] Within the potential grouting influence domain, a multi-level cascaded numerical simulation is performed to generate an initial grouting strategy set; the initial grouting strategy set includes a grouting pressure adjustment scheme, a grouting material ratio scheme, and a grouting path planning scheme.

[0013] Receive external input signals, determine the correction operation type based on the external input signals, and dynamically correct the initial grouting strategy set based on the correction operation type;

[0014] The initial grouting strategy set is dynamically modified, and the numerical map model is updated based on the execution feedback data.

[0015] Preferably, obtaining the real-time state parameters of the material layer includes:

[0016] Collect in-situ soil sample data of the material layer;

[0017] The moisture content, void ratio, and mineral composition percentage of the in-situ soil sample data were determined.

[0018] Acquire real-time environmental monitoring data of the material layer, including groundwater level fluctuations and soil temperature gradient changes.

[0019] By integrating the moisture content, void ratio, mineral composition ratio, groundwater level fluctuation value, and soil temperature gradient change value, the real-time state parameters of the material layer are generated.

[0020] Preferably, calculating the material state feature vector of the material layer includes:

[0021] The real-time state parameters of the material layer are input into a pre-trained material property mapping model;

[0022] The material property mapping model maps the real-time state parameters of the material layer to a vector space of fixed dimension.

[0023] The output mapped vector is used as the material state feature vector, wherein the material state feature vector characterizes the comprehensive physical and mechanical properties of the material layer.

[0024] Preferably, the step of dynamically locating the potential grouting influence domain of the fluid layer in the numerical mapping model using a multi-layer progressive calculation method based on the material state feature vector includes:

[0025] The material state feature vector is input into the numerical mapping unit;

[0026] The material state feature vector is mapped to the node attribute space of the numerical graph model through the numerical mapping unit.

[0027] Calculate the similarity between the mapped material state feature vector and the node attribute vectors of all analyzed nodes in the numerical graph model;

[0028] Select analysis nodes whose similarity reaches a preset threshold as initial influence nodes;

[0029] Starting from the initial influencing node, multi-layer progressive diffusion calculations are performed outward along the associated edge;

[0030] In each diffusion calculation, it is determined whether the correlation strength value between the node attribute vector of the newly touched analysis node and the material state feature vector exceeds the diffusion threshold.

[0031] If the diffusion threshold is exceeded, the newly reached analysis node will be included in the diffusion set;

[0032] Diffusion stops when the preset number of diffusion layers is reached or the correlation strength value is lower than the diffusion threshold.

[0033] The spatial extent corresponding to the final diffusion set is defined as the potential grouting influence domain.

[0034] Preferably, the step of performing multi-level cascaded numerical simulations to generate an initial grouting strategy set includes:

[0035] Within the potential grouting influence zone, extract the structural stress distribution data of the structural layer;

[0036] Based on the structural stress distribution data, the equivalent permeability tensor of the fluid layer is calculated;

[0037] Based on the equivalent permeability tensor, the diffusion path of grout in the fluid layer under different grouting pressures is simulated.

[0038] Based on the diffusion path simulation results, the grouting path planning scheme is generated;

[0039] Simultaneously analyze the matching degree between the diffusion path and the material parameters of the material layer to generate the grouting material ratio scheme;

[0040] The grouting pressure adjustment scheme is generated by combining the equivalent permeability tensor and the preset safety threshold.

[0041] Preferably, the step of dynamically modifying the initial grouting strategy set based on the modification operation type includes:

[0042] If the correction operation type is material property correction, then the material state feature vector is recalculated and the grouting material ratio scheme is updated;

[0043] If the correction operation type is structural constraint correction, then the structural stress distribution data is re-extracted and the grouting path planning scheme is updated;

[0044] If the correction operation type is fluid parameter correction, then the equivalent permeability tensor is recalculated, and the grouting pressure adjustment scheme and the grouting path planning scheme are corrected simultaneously.

[0045] Preferably, updating the numerical graph model based on execution feedback data includes:

[0046] Collect real-time monitoring data during the grouting construction process. The real-time monitoring data includes the actual diffusion radius of the grout, the grouting pressure fluctuation value, and the formation uplift.

[0047] The fluid permeability coefficient of the fluid layer is updated based on the deviation between the actual diffusion radius of the slurry and the simulated diffusion path.

[0048] The structural stress state of the structural layer is updated based on the degree of deviation between the grouting pressure fluctuation value and the preset pressure threshold.

[0049] The material parameter variation values ​​of the material layer are inverted based on the formation uplift.

[0050] Based on the updated fluid permeability coefficient, structural stress state, and material parameter changes, the dynamic attribute set of the relevant analysis nodes in the numerical graph model is recalculated.

[0051] Preferably, the method for calculating the correlation strength value includes:

[0052] Extract the historical state evolution records of the target analysis nodes in the numerical mapping model;

[0053] Calculate the temporal correlation index between the historical state evolution record and the material state feature vector;

[0054] Obtain the attribute coupling degree between the target analysis node and the upstream analysis node;

[0055] The temporal correlation index and the attribute coupling degree are fused according to a preset ratio to generate the correlation strength value.

[0056] Preferably, calculating the equivalent permeability tensor of the fluid layer includes:

[0057] Extract the principal stress directions and stress gradient values ​​from the structural stress distribution data;

[0058] The permeation anisotropy principal axis is determined based on the principal stress direction;

[0059] The permeability correction factor for each principal axis direction is calculated based on the stress gradient value.

[0060] The component values ​​of the equivalent permeability tensor are obtained by multiplying the reference permeability of the fluid layer by the permeability correction factor in the corresponding direction.

[0061] Preferably, the step of inverting the material parameter variation value of the material layer based on the formation uplift includes:

[0062] Establish the inverse mapping relationship between the formation uplift and the soil compression modulus;

[0063] Based on the inversion mapping relationship and the measured formation uplift, the updated value of the compressive modulus of the material layer is calculated;

[0064] Based on the updated value of the compression modulus, the porosity and moisture content parameters of the material layer are corrected in reverse.

[0065] Compared with the prior art, the beneficial effects of the present invention are:

[0066] This invention significantly improves the systematicness and hierarchy of grouting reinforcement simulation by establishing a three-layer deconstruction architecture of building foundations—material layer, structural layer, and fluid layer—and constructing a numerical atlas model integrating multi-node attributes and relationships. By acquiring material state parameters in real time and calculating eigenvectors, it achieves coherent deduction from material properties to fluid permeation behavior, enhancing the model's ability to characterize the actual state of the foundation. Utilizing a multi-layered progressive calculation mechanism, the potential impact range of grouting operations can be dynamically identified, avoiding simulation biases caused by simplification assumptions in traditional methods.

[0067] Multi-level cascaded numerical simulations are conducted within the potential grouting influence domain to simultaneously generate various strategy schemes for pressure adjustment, material proportioning, and path planning, forming a systematic initial strategy set. This method overcomes the limitations of single-parameter optimization, achieving multi-factor collaborative design and making the grouting strategy more comprehensive and reasonable. By introducing an external input signal to trigger a dynamic correction mechanism, the model becomes open and adaptable, capable of responding to environmental changes or human intervention and continuously adjusting the strategy content.

[0068] By implementing corrective strategies and updating numerical maps based on feedback, a closed-loop learning process of "simulation-execution-update" is formed, enabling the model to continuously evolve. This mechanism significantly improves the effectiveness and robustness of grouting strategies, especially demonstrating a clear advantage when dealing with foundation conditions characterized by high uncertainty and significant time-varying features. The entire method constructs a complete technical chain from state awareness and dynamic computation to strategy generation and optimization, providing an efficient and reliable digital simulation tool for complex foundation repair. Attached Figure Description

[0069] Figure 1 This is a schematic diagram illustrating the working principle of the numerical simulation method for grouting reinforcement and repair of building foundations as described in this invention.

[0070] Figure 2 A flowchart for calculating the eigenvectors of material states;

[0071] Figure 3 A flowchart for dynamically locating potential grouting influence areas;

[0072] Figure 4 A flowchart for generating the initial grouting strategy set;

[0073] Figure 5 A flowchart for dynamically modifying the initial grouting strategy set. Detailed Implementation

[0074] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0075] Please see Figure 1 This invention provides a numerical simulation method for grouting reinforcement and repair of building foundations, the overall implementation of which includes the following steps:

[0076] The building foundation to be repaired is deconstructed into three layers: a material layer, a structural layer, and a fluid layer. A numerical mapping model is established, which includes multiple analysis nodes, associated edges connecting the analysis nodes, and dynamic attribute sets for each analysis node. The dynamic attribute sets include material parameters, structural stress state, and fluid permeability coefficient. Real-time state parameters of the material layer are obtained, and a material state feature vector of the material layer is calculated based on these parameters. Based on the material state feature vector, the potential grouting influence domain of the fluid layer is dynamically located in the numerical mapping model using a multi-layered progressive calculation method. Multi-layered cascaded numerical simulations are performed within the potential grouting influence domain to generate an initial grouting strategy set, which includes grouting pressure adjustment schemes, grouting material proportioning schemes, and grouting path planning schemes. External input signals are received, and the correction operation type is determined based on these signals. The initial grouting strategy set is dynamically corrected based on the correction operation type. The dynamically corrected initial grouting strategy set is executed, and the numerical mapping model is updated based on the execution feedback data.

[0077] The system hardware architecture includes a data acquisition layer, comprising in-situ sensors (water content sensor, stress sensor, permeability coefficient tester) and environmental monitoring equipment (groundwater level gauge, temperature sensor), with a data sampling frequency of 1 time / minute and a sampling accuracy of ≤±2%. The computing and processing layer utilizes an industrial-grade server (CPU, memory ≥64GB, hard disk ≥2TB), supporting parallel computing to meet the real-time requirements of multi-layer cascaded numerical simulations. The output control layer connects to the grouting equipment controller and the data storage server, enabling real-time issuance of strategy commands and synchronous storage of feedback data.

[0078] The software architecture includes a data processing module developed in Python, integrating data cleaning, interpolation calculation, and feature extraction functions, and supporting the import and integration of multi-source data formats (CSV, JSON). A numerical simulation module, with a core computing engine developed in C++, encapsulates multi-layer progressive diffusion algorithms, equivalent penetration tensor calculation algorithms, etc., and provides API interfaces for interaction with other modules. A policy generation and correction module, developed in C#, includes signal analysis, policy adjustment, and logging submodules, supporting modular calls and functional expansion. A model update module integrates database interaction functions, reads feedback data in real time, automatically triggers model attribute updates, and maintains an update frequency consistent with the feedback data acquisition frequency. The system debugging process includes unit testing, independently testing each module to verify data processing accuracy, algorithm calculation accuracy, and command issuance timeliness. Integration testing simulates actual engineering scenarios to test the stability of data transmission between modules and the smoothness of logical connections. Field trial operation was conducted in a small-scale foundation repair project for 72 hours, recording system operating parameters and optimizing response speed and stability.

[0079] Example 1: The specific construction process of the numerical mapping model is as follows: Node division rules are based on the three-dimensional spatial grid of the foundation. The grid edge length is set according to the foundation size and accuracy requirements, with a range of 0.5m-2m, ensuring that the nodes cover the entire area of ​​the material layer, structural layer, and fluid layer. The logic for constructing associated edges includes spatially associated edges, establishing physical connections between adjacent grid nodes. The initial edge weight is calculated based on the spatial distance between nodes, with closer nodes having higher weights. In attribute-associated edges, when the similarity of the dynamic attributes (material parameters, stress state, permeability coefficient) of two nodes is ≥0.7, an attribute-associated edge is established, with the edge weight positively correlated with the similarity. The material parameters for initializing the dynamic attribute set are obtained through in-situ tests, including the soil's natural density, shear strength, and compression modulus, and are filled into the corresponding material layer nodes. The structural stress state is calculated using finite element software to determine the initial stress state of the foundation, and the magnitude and direction of the principal stresses at each node are entered. The fluid permeability coefficient is determined by combining indoor permeability test results with regional geological survey data to determine the initial permeability coefficient of each fluid layer node. During model verification and optimization, the initial parameters are substituted into the model for trial calculations. The calculation results are compared with the actual measured data on site. If the deviation exceeds 10%, the node weights or initial attribute values ​​are adjusted until the deviation meets the engineering accuracy requirements.

[0080] See Figure 2 This embodiment relates to the specific process of acquiring real-time state parameters of a material layer and calculating its characteristic vector. This process begins with comprehensive data acquisition and processing of the building foundation material layer. The acquisition of real-time state parameters of the material layer is first achieved by collecting in-situ soil sample data. The in-situ soil sample data is collected using standard geological drilling methods, employing a hollow tube drill to obtain undisturbed soil samples at different depths. Sampling points are distributed in a grid pattern according to the foundation plane, with samples taken at certain intervals along the depth direction to ensure spatial representativeness of the data. The obtained soil samples are immediately sealed and sent to the laboratory for index determination. The determination process includes using the oven drying method to determine the moisture content, calculating the porosity using the hydrometer method and volume measurement, and determining the mineral composition percentage using X-ray diffraction analysis. These determination methods comply with current geotechnical testing standards, ensuring the accuracy and comparability of the data.

[0081] While acquiring laboratory measurement data, real-time environmental monitoring data of the material layer is obtained through a monitoring system deployed on-site. The monitoring system includes groundwater level monitoring wells and a temperature sensor array. Groundwater level monitoring wells are arranged around the building foundation and in key areas, with pressure-type water level sensors installed inside to record water level changes at a certain frequency and calculate groundwater level fluctuations. The temperature sensor array is buried in the soil layer in a three-dimensional grid format. The sensors use thermocouples or resistance thermometers to record temperature values ​​at different depths, and the soil temperature gradient is calculated through spatial interpolation. All sensor data is automatically recorded by a data acquisition unit and transmitted to the central processing unit.

[0082] Laboratory measurement data and real-time monitoring data are integrated into a unified data processing platform. The platform performs time synchronization and spatial alignment processing on data from different sources. Moisture content, porosity, and mineral composition data are correlated and mapped based on the spatial relationship between their sampling locations and monitoring points. For soil sample data located between monitoring points, Kriging interpolation is used to generate continuous spatial distribution data. Groundwater level fluctuations and soil temperature gradient changes are directly correlated with soil sample data within their influence range. The integration process considers the physical relationships between various parameters, such as correlating water level fluctuations with porosity changes and temperature gradients with mineral activity. The final generated real-time state parameters of the material layer are a multi-dimensional dataset containing spatial coordinate information, timestamps, and numerical values ​​for each parameter.

[0083] When calculating the material state feature vector, the real-time state parameters of the material layer are input into a pre-trained material property mapping model. This model is trained based on a large amount of historical soil engineering data, covering various soil types, environmental conditions, and mechanical states. The model employs a deep neural network architecture, containing multiple hidden layers and nonlinear activation functions. The input layer dimension is consistent with the dimension of the real-time state parameters of the material layer, receiving parameters such as water content, porosity, mineral composition ratio, groundwater level fluctuation, and soil temperature gradient. Internally, the model uses forward propagation to extract features layer by layer and perform nonlinear transformations. The first hidden layer performs linear combination and normalization of the input parameters, and subsequent hidden layers progressively abstract higher-order feature representations. The output layer projects the final features onto a fixed-dimensional vector space, generating a material state feature vector.

[0084] The model training process employs supervised learning, with training labels consisting of mechanical property parameters of the corresponding soil samples, such as shear strength and compression modulus. A loss function measures the difference between predicted features and actual mechanical properties, and the model weights are optimized using a backpropagation algorithm. The trained model can map complex multi-parameter inputs into a compact feature representation, which fully captures the comprehensive physical and mechanical properties of the soil material. In practical applications, the calculation of the material state feature vector is completely automated, requiring no manual intervention. The calculated feature vector is stored as a numerical array, with each dimension representing a specific abstract material property, such as impermeability, deformation characteristics, or strength properties. This vector serves as a digital representation of the material layer, providing standardized input for subsequent analysis.

[0085] In-situ sampling follows random sampling principles to ensure sample representativeness; laboratory measurements employ standard operating procedures to minimize human error; the monitoring system layout considers both spatial coverage and measurement accuracy; data processing utilizes mature statistical and interpolation methods; and the feature mapping model has undergone thorough validation and testing. Through this series of rigorous steps, the resulting material state feature vector accurately reflects the actual state of the material layer. All data processing and calculations are automated within the computer system, ensuring efficiency and consistency.

[0086] Example 2: See Figure 3 This embodiment illustrates a method for dynamically locating the potential grouting influence domain of a fluid layer in a numerical mapping model based on material state feature vectors, and for calculating the associated strength values. The process begins by inputting the material state feature vectors into a numerical mapping unit. The numerical mapping unit is a computational module that transforms the input feature vectors into the node attribute space of the numerical mapping model. The node attribute space is a multi-dimensional space, where each dimension corresponds to a specific attribute feature of the dynamic attribute set of the analysis node. The mapping process is achieved through a linear transformation, projecting each component of the material state feature vector onto the basis vectors of the node attribute space to generate a new vector representation with the same dimension and scale as the node attribute vectors.

[0087] After mapping, the system calculates the similarity between the mapped material state feature vector and the node attribute vectors of all analyzed nodes in the numerical graph model. The similarity calculation uses a cosine similarity algorithm, evaluating directional consistency by calculating the cosine of the angle between two vectors in space. The calculation process traverses each analyzed node in the numerical graph model, generating a corresponding similarity value. These values ​​are normalized and compared with a preset threshold. The preset threshold is set based on historical engineering data, reflecting the minimum correlation required between material properties and node attributes. Based on statistical analysis of 100 sets of foundation grouting test data under different geological conditions, when the similarity is ≥0.7, the matching degree between node attributes and material state features meets the positioning accuracy requirements of the grouting influence domain; below this value, positioning deviations are likely to occur. Combining soil mechanical properties and grout diffusion laws, when the correlation strength value is ≥0.5, the foundation area corresponding to the node is significantly affected by grouting; below this value, the grouting effect on improving foundation performance is weak, and it is unnecessary to include it in the influence domain. All analyzed nodes with similarity reaching or exceeding this threshold are marked as initial influence nodes.

[0088] Starting from the initial influencing node, the system performs multi-layered progressive diffusion calculations outward along the associated edges. In the numerical graph model, associated edges represent the connections between nodes, and each edge has a weight attribute reflecting the strength of the mutual influence between nodes. The diffusion calculation employs a breadth-first search algorithm, starting from the initial set of influencing nodes and visiting its connected neighboring nodes layer by layer. In each layer of diffusion calculation, the association strength value is calculated for newly reached analysis nodes. The association strength value quantifies the degree of association between the new node and the material state eigenvector.

[0089] The calculation of the correlation strength value involves several steps. First, the historical state evolution records of the target analysis node are extracted from the numerical mapping model. These historical state evolution records are time-series data retrieved from the model database, recording the dynamic attribute set values ​​of the node at multiple past time points. This data includes the changes in material parameters, structural stress state, and fluid permeability coefficient over time. Next, the temporal correlation index between the historical state evolution records and the material state eigenvectors is calculated. This index is calculated using a dynamic time warping algorithm to assess the similarity between the two time series in terms of shape and change patterns, and can capture nonlinear temporal relationships.

[0090] Simultaneously, the system acquires the attribute coupling degree between the target analysis node and upstream analysis nodes. Upstream analysis nodes refer to nodes that have been visited during the diffusion process and are directly connected to the current target node. Attribute coupling degree is determined by calculating the Euclidean distance between the dynamic attribute sets of the two nodes and combining it with the weight values ​​of the connecting edges. The smaller the distance and the larger the edge weight, the higher the attribute coupling degree. After the temporal relevance index and attribute coupling degree are calculated, they are fused according to a preset ratio. The preset ratio is based on domain knowledge and typically assigns higher weight to temporal relevance because historical evolution patterns better reflect the behavioral characteristics of nodes. The fused values ​​are then standardized to generate the final association strength value.

[0091] In each diffusion calculation, the newly calculated association strength value is compared with the diffusion threshold. The diffusion threshold is a dynamically adjusted parameter, with its initial value set based on engineering experience and adaptively adjusted according to the attribute characteristics of visited nodes during the diffusion process. If the association strength value exceeds the diffusion threshold, the currently analyzed node is included in the diffusion set and marked as the starting point for the next diffusion layer. The diffusion process continues until the preset maximum number of diffusion layers is reached, or the association strength values ​​of newly visited nodes are generally lower than the diffusion threshold. The preset number of diffusion layers is determined according to the project scale and accuracy requirements, and is usually set to a depth of several layers. For common foundation thicknesses (3m-20m), 3-5 diffusion layers can cover the actual impact range of grouting. Too few layers may miss key areas, while too many layers will lead to calculation redundancy.

[0092] When the diffusion process terminates, the system merges the spatial extents corresponding to all analyzed nodes in the diffusion set to form the final potential grouting influence domain. This influence domain is a three-dimensional spatial region determined by the geographical coordinates represented by all relevant nodes. The system outputs information such as the boundary coordinates, spatial morphology, and node list of the influence domain, providing spatial extent definition for subsequent grouting strategy generation.

[0093] The entire implementation process is automated within a computer system. The numerical graph model is stored in a graph database, supporting efficient node traversal and relationship lookup. Diffusion computation employs parallel computing technology to accelerate processing, especially for large-scale numerical graph models, where a distributed computing framework distributes computational tasks across multiple processing units. All intermediate computation results, including similarity values, association strength values, and diffusion sets, are stored in real-time in an in-memory database for use in subsequent steps. Various thresholds and parameters involved in the implementation process are managed through configuration files, allowing for adjustments based on specific engineering conditions. This implementation method, through a systematic computational process, achieves accurate identification of potential grouting influence domains, providing a spatial basis for formulating grouting reinforcement strategies.

[0094] Example 3: See Figure 4 This embodiment relates to the process of performing multi-level cascaded numerical simulations within a potential grouting influence zone to generate an initial set of grouting strategies. The process begins with extracting structural stress distribution data from the numerical mapping model. This data includes the magnitude and direction of principal stresses at each spatial location, stress gradient values, and the components of the stress tensor. These data are obtained through finite element analysis and reflect the mechanical response state of the foundation under the current loading conditions. The data extraction scope is strictly limited to the determined potential grouting influence zone to ensure the relevance and effectiveness of subsequent calculations.

[0095] Based on the acquired structural stress distribution data, the system calculates the equivalent permeability tensor of the fluid layer. The calculation process first identifies the principal stress directions, determining three mutually perpendicular principal stress axes using eigenvalue decomposition. These principal axes define the main directions of permeability anisotropy. Subsequently, permeability correction factors for each principal axis direction are calculated based on the stress gradient values. The stress gradient values, reflecting the severity of stress changes, are calculated using the spatial difference method. The permeability correction factors characterize the enhancing or weakening effect of the stress state on the original permeability.

[0096] The equivalent permeability tensor is calculated using the following relationship:

[0097]

[0098] in: The components of the equivalent permeability tensor are represented. The baseline permeability of the fluid layer is represented. This represents the penetration rate correction factor. The stress gradient value is represented. The baseline permeability is obtained from the dynamic property set of the fluid layer and represents the inherent permeability characteristics under stress-free conditions. The permeability correction factor is a function of the stress gradient value, determined through empirical relationships, and reflects the influence of the stress field on pore structure and connectivity. The calculated equivalent permeability tensor fully describes the permeability characteristics of the fluid layer under the current stress state.

[0099] Based on the equivalent permeability tensor, the system simulates the diffusion path of grout in the fluid layer under different grouting pressures. The simulation process is based on porous media fluid mechanics theory, using the finite element method to solve the governing equations of grout flow. The computational domain is discretized into mesh elements, each assigned a corresponding equivalent permeability tensor value. The simulation considers the rheological properties of the grout, including viscosity changes over time. For each set grouting pressure value, the system calculates the advancement process of the grout front, recording the diffusion range, path morphology, and filling degree. The simulation results generate a series of pressure-diffusion relationship curves, showing the grout distribution characteristics under different pressure conditions.

[0100] Based on the diffusion path simulation results, the system generates a grouting path planning scheme. This scheme determines the location, sequence, and depth of the grouting holes. The locations are chosen in areas with poor diffusion or stress concentration. The grouting sequence follows the principle of from the periphery to the center and from bottom to top. The grouting depth is determined based on the vertical distribution characteristics of the grout in the diffusion simulation. The scheme also includes the technical parameters for each grouting hole, including borehole diameter, casing configuration, and sealing requirements.

[0101] The matching degree between the diffusion path and the material parameters of the material layer is analyzed simultaneously to generate a grouting material mix design. The matching degree analysis includes the adaptability of the grout viscosity to the soil pore size, the coordination of the grout setting time to the groundwater flow rate, and the compliance of the grout strength with the foundation bearing capacity requirements. Based on the analysis results, the water-cement ratio, additive type and dosage, and gel time control parameters of the grout are determined. The mix design may vary for different grouting areas, reflecting the zoning and classification design concept of grouting materials.

[0102] A grouting pressure adjustment scheme is generated by combining the equivalent permeability tensor and a preset safety threshold. The safety threshold is determined based on the formation's resistance to fracturing and the protection requirements of adjacent structures. The pressure adjustment scheme specifies the initial grouting pressure, pressure ramp-up rate, maximum allowable pressure, and pressure control strategy. For areas with poor permeability, the scheme may recommend a staged pressure ramp-up method; for sensitive areas, the scheme specifies pressure monitoring values ​​and alarm thresholds. The scheme also includes the coordination between pressure and flow rate to ensure the safety and effectiveness of the grouting process.

[0103] The entire implementation process was completed within a numerical simulation platform, which integrates functional modules such as geological modeling, mechanical analysis, fluid simulation, and optimization algorithms. The calculations employed an iterative approach, first simulating based on initial parameters, then adjusting the calculation parameters according to intermediate results, and finally outputting a stable solution. All generated solution data was stored in the project database, including simulation input parameters, intermediate results, and the final recommended solution. Solution output formats included data tables, graphs, and 3D visualizations, supporting engineers in reviewing solutions and making optimization decisions. A complete calculation log was maintained throughout the implementation process, recording parameter settings and result outputs for each calculation step, ensuring traceability and repeatability.

[0104] Example 4: See Figure 5 This embodiment relates to the process of dynamically correcting an initial grouting strategy set based on external input signals. Taking a high-rise building foundation grouting reinforcement project located on a river alluvial plain as an example, during the implementation of the initial grouting strategy, the automated monitoring system discovered a difference between the actual grouting effect and the expected result. The real-time data collected by the monitoring system is transmitted to the processing system as external input signals. The system first analyzes the characteristic patterns of these signals to identify the type of correction operation that needs to be performed.

[0105] The signal analysis object is the external input signal, including monitoring data signal and manual command signal. It extracts the numerical change and rate of change in the monitoring data signal, compares it with the preset threshold, and determines whether it belongs to the change of material properties, structural constraints or fluid parameters. It identifies the keywords in the manual command signal and matches the corresponding correction operation type.

[0106] In this specific example, monitoring data showed that the soil moisture content in the grouting area increased by approximately 15% compared to the initial survey, while mineral composition analysis revealed an increase in clay mineral content. The system identified these changes as alterations to material properties and therefore determined the correction operation type to be material property correction. The system immediately initiated a process to reacquire real-time state parameters of the material layer, collecting the latest moisture content and mineral composition data through a sensor network deployed in the grouting area, while simultaneously conducting laboratory verification using borehole sampling. After integration and processing, the reacquired data was input into a pre-trained material property mapping model to generate an updated material state feature vector. This new feature vector reflects the current actual state of the soil material, exhibiting higher plasticity and lower permeability compared to the initial state.

[0107] Based on the updated material state feature vector, the system adjusts the grouting material mix design accordingly. The material property correction, combined with the updated material state feature vector, queries a preset material parameter-grouting material mix mapping table to automatically adjust the proportions of cementitious materials, aggregates, and admixtures in the grout, ensuring the mix design matches the current material layer characteristics. The original cement-water glass two-component grout mix was modified to a ternary system grout with added bentonite, the water-cement ratio was adjusted from the initial 0.8 to 1.0, and the gel time was extended from 3 minutes to 5 minutes. These adjustments aim to adapt to the changed soil conditions, ensuring the grout can fully penetrate and effectively solidify. The modification of the mix design not only involves adjusting the proportions of the main materials but also optimizing the types and amounts of admixtures, such as increasing the amount of retarder to cope with higher moisture content conditions.

[0108] In another scenario, the monitoring system detected minor displacement of the adjacent diaphragm wall during grouting. Although the displacement did not exceed the safety threshold, it showed a continuous development trend. The system identified this as a need for structural constraint correction and immediately initiated the re-extraction of structural stress distribution data. The current stress state was recalculated using finite element analysis software, obtaining updated stress distribution cloud maps and deformation field data. The new analysis results showed that, due to the grouting pressure, a previously unconsidered weak interlayer experienced stress concentration.

[0109] Based on the re-extracted structural stress distribution data, the system updates the grouting path planning scheme. After re-extracting the structural stress distribution data and incorporating structural constraint corrections, stress concentration areas and deformation-sensitive areas are identified. When adjusting the grouting path, stress concentration areas are avoided to prevent further structural damage, while the number of grouting holes is increased in deformation-sensitive areas to enhance the reinforcement effect. The originally designed radial grouting path has been modified to a layered and segmented grouting scheme, with the grouting sequence adjusted to first reinforce the weak interlayer before grouting the main body area. The spacing between grouting holes has been adjusted from 1.5 meters to 1.2 meters, and the grouting depth has been increased by 2 meters in the weak interlayer area. The path planning also incorporates targeted protection measures for adjacent continuous walls, including the installation of stress relief holes and monitoring sections. The new path planning scheme ensures the uniformity of stress distribution during grouting and avoids adverse effects on surrounding structures.

[0110] Another scenario is that the monitoring system detects a significant deviation between the actual diffusion range of the slurry and the simulation results, with the diffusion radius in a certain area being 30% smaller than expected. System analysis indicates a change in fluid parameters, therefore the correction operation type is determined to be fluid parameter correction. The system recalculates the equivalent permeability tensor for that area and updates the permeability coefficient model based on the latest pressurized water test data and pore water pressure monitoring results.

[0111] Based on the recalculated equivalent permeability tensor, the system simultaneously corrects the grouting pressure adjustment scheme and the grouting path planning scheme. After recalculating the equivalent permeability tensor in conjunction with fluid parameter corrections, and considering the grout rheological properties, the grouting pressure ramp-up rate is adjusted. For areas with reduced permeability, a low initial pressure and gradual pressure ramp-up mode are adopted to avoid formation damage caused by sudden pressure increases. The upper limit of the grouting pressure is increased from the initial 2.0 MPa to 2.5 MPa, and the pressure ramp-up gradient is adjusted from 0.2 MPa per stage to 0.3 MPa. Simultaneously, supplementary grouting holes are added in areas with insufficient diffusion, the grouting rhythm is adjusted, and an intermittent grouting method is adopted to improve grout permeability. Regarding path planning, borehole inclination control requirements are added to ensure that the grouting holes accurately reach the target strata.

[0112] Throughout the entire remediation process, the system maintains version records and explanations of the reasons for all modifications. Each remediation operation generates a detailed remediation report, including the original parameters, the parameters after remediation, the basis for the remediation, and an assessment of the expected effects. The system also establishes a predictive model for the remediation effects, conducting forward-looking analyses of the potential impact of each remediation to provide a reference for engineers' decision-making.

[0113] The system employs a modular correction strategy, allowing different types of correction operations to be performed independently or in combination. All correction operations are completed on a unified control platform, ensuring coordination and consistency among the various correction items. The modified grouting strategy set is only applied to actual grouting operations after multiple rounds of simulation verification, guaranteeing the safety and reliability of the project. This dynamic correction mechanism enables the grouting strategy to adapt to real-time changes in foundation conditions, ensuring that the grouting effect achieves the expected goals through continuous optimization.

[0114] Example 5: Updating the numerical mapping model based on execution feedback data. This process begins with the acquisition of real-time monitoring data during grouting construction. The monitoring system is deployed within the grouting influence area and includes grout pressure sensors, displacement monitoring points, and ground deformation measurement devices. The actual grout diffusion radius is measured using resistivity tomography systems deployed at different borehole depths. This system tracks the diffusion front by detecting the difference in conductivity between the grout and the surrounding soil. Grouting pressure fluctuations are recorded by pressure sensors installed in the grouting pipeline, with a sampling frequency of several times per second, capable of capturing instantaneous pressure changes. Based on geotechnical engineering safety specifications, the grouting pressure does not exceed 80% of the soil's splitting strength, preventing the generation of new foundation cracks during grouting while ensuring effective grout diffusion. Ground uplift is monitored jointly by precision leveling and a hydrostatic level installed on the ground surface, achieving sub-millimeter accuracy.

[0115] The collected real-time monitoring data, after preprocessing, is used to update the numerical mapping model. Based on the deviation between the actual diffusion radius and the simulated diffusion path of the slurry, the system updates the fluid permeability coefficient of the fluid layer. Deviation analysis employs spatial statistical methods to calculate the overlap index between the actual diffusion range and the simulated range. For areas with large deviations, the permeability coefficient value is adjusted using back analysis. The adjustment process considers the interaction between slurry properties and soil structure, employing an iterative algorithm to gradually correct the permeability coefficient until the simulation results and actual observation data achieve the required level of agreement.

[0116] The system updates the structural stress state of the structural layer based on the deviation of the grouting pressure fluctuation value from the preset pressure threshold. The pressure fluctuation analysis employs a time-frequency analysis method to identify characteristic patterns in the pressure sequence. For areas exhibiting abnormal pressure fluctuations, the stress distribution is recalculated using a coupled fluid-solid mechanics model. The stress update considers the hysteresis effect between grouting pressure and formation response, employing a viscoelastic constitutive model to describe the time-dependent behavior of the soil. The updated stress state reflects the actual mechanical response that occurred during the grouting process.

[0117] The material parameter variations of the material layer are derived from the formation uplift. The inversion process is based on the mapping relationship between formation uplift and soil compression modulus, which is described by the following expression:

[0118]

[0119] in: The soil compression modulus is a mechanical parameter that reflects the soil's ability to resist compressive deformation, and its unit is megapascal (MPa). The empirical coefficient 1 is a dimensionless coefficient obtained through regression analysis of historical engineering data, reflecting the proportional relationship between the amount of formation uplift and the change in the compression modulus. It represents the vertical displacement of a point on the ground surface or inside the stratum during the grouting process, and is expressed in millimeters (mm). This indicates the initial stratum thickness, which is the original thickness of the target soil layer before grouting treatment, in meters (m). The second empirical coefficient is a coefficient (MPa) with stress units obtained through regression analysis of historical engineering data, reflecting the initial compression characteristics of the soil. Empirical coefficients are determined through laboratory tests and field data regression analysis, reflecting the mechanical properties of specific soil types. Based on measured ground uplift, the updated value of the material layer's compression modulus is calculated using this formula.

[0120] Based on the updated compression modulus value, the system reverse-corrects the porosity and moisture content parameters of the material layer. The correction process employs a soil constitutive model, considering the empirical relationship between compression modulus and porosity. For saturated soil, a quantitative relationship between the change in porosity and compression modulus is derived based on the effective stress principle and compression characteristics. Moisture content correction is based on the soil-water characteristic curve, considering the impact of water expulsion during compression. All parameter corrections adhere to the principles of mass and energy conservation to ensure the rationality of the physical process.

[0121] The updated fluid permeability coefficient, structural stress state, and material parameter changes were used to recalculate the dynamic attribute sets of relevant analysis nodes in the numerical spectroscopic model. The update process employed an incremental update algorithm, modifying only the attribute values ​​of nodes that had changed, thus improving computational efficiency. Each node's update record included the modification time, reason for modification, and a comparison of values ​​before and after the modification, forming a complete modification log.

[0122] The model update log records the following: the reason for the update, the specific parameters of the update, and a comparison of the model prediction results before and after the update. The log provides data support for subsequent grouting strategy optimization. When this method is reused in projects with similar geological conditions, initial parameters can be quickly set based on historical logs.

[0123] The system establishes a feedback control loop that periodically compares the differences between monitoring data and model predictions. When the difference exceeds a set tolerance, the model update process is automatically triggered. The updated model is immediately used for subsequent grouting process simulation and strategy optimization, forming a closed-loop control system of monitoring-update-application.

[0124] The entire implementation process runs on a distributed computing platform capable of handling large-scale monitoring data and high-precision numerical models. The data acquisition, processing, and analysis modules adopt a microservice architecture, with communication between modules via standard interfaces. The model update algorithm has been optimized to improve computational speed while maintaining accuracy, meeting the real-time requirements of engineering implementation.

[0125] A rigorous quality control procedure was established during implementation. All monitoring data underwent outlier detection and reliability verification, and model update results required cross-validation before being applied to actual projects. The system is also equipped with visualization tools that can intuitively display changes before and after model updates, helping engineers understand the foundation response mechanism and grouting effect.

[0126] This model update mechanism, based on execution feedback data, enables the numerical mapping model to continuously reflect the actual state of the foundation, providing an accurate numerical simulation basis for grouting reinforcement. By continuously incorporating field monitoring data, the model gradually improves its predictive capabilities, forming an increasingly accurate foundation response prediction capability.

[0127] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0128] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A numerical simulation method for grouting reinforcement and repair of building foundations, characterized in that, include: The building foundation to be repaired is deconstructed into three levels; wherein, the levels include a material layer, a structural layer, and a fluid layer; A numerical graph model is established, which includes multiple analysis nodes, associated edges connecting the analysis nodes, and a dynamic attribute set for each analysis node. The dynamic attribute set includes material parameters, structural stress state, and fluid permeability coefficient. Obtain the real-time state parameters of the material layer; Based on the real-time state parameters of the material layer, calculate the material state feature vector of the material layer; Based on the material state feature vector, the potential grouting influence domain of the fluid layer is dynamically located in the numerical mapping model using a multi-layer progressive calculation method, including: The material state feature vector is input into the numerical mapping unit; The material state feature vector is mapped to the node attribute space of the numerical graph model through the numerical mapping unit. Calculate the similarity between the mapped material state feature vector and the node attribute vectors of all analyzed nodes in the numerical graph model; Select analysis nodes whose similarity reaches a preset threshold as initial influence nodes; Starting from the initial influencing node, multi-layer progressive diffusion calculations are performed outward along the associated edge; In each diffusion calculation, it is determined whether the correlation strength value between the node attribute vector of the newly touched analysis node and the material state feature vector exceeds the diffusion threshold. If the diffusion threshold is exceeded, the newly reached analysis node will be included in the diffusion set; Diffusion stops when the preset number of diffusion layers is reached or the correlation strength value is lower than the diffusion threshold. The spatial range corresponding to the final diffusion set is defined as the potential grouting influence domain; The method for calculating the correlation strength value includes: Extract the historical state evolution records of the target analysis nodes in the numerical mapping model; Calculate the temporal correlation index between the historical state evolution record and the material state feature vector; Obtain the attribute coupling degree between the target analysis node and the upstream analysis node; The time-series correlation index and the attribute coupling degree are fused according to a preset ratio to generate the correlation strength value; Within the potential grouting influence domain, multi-level cascaded numerical simulations are performed to generate an initial set of grouting strategies, including: Within the potential grouting influence zone, extract the structural stress distribution data of the structural layer; Based on the structural stress distribution data, the equivalent permeability tensor of the fluid layer is calculated; Based on the equivalent permeability tensor, the diffusion path of grout in the fluid layer under different grouting pressures is simulated. Based on the diffusion path simulation results, the grouting path planning scheme is generated; Simultaneously analyze the matching degree between the diffusion path and the material parameters of the material layer to generate a grouting material ratio scheme; The grouting pressure adjustment scheme is generated by combining the equivalent permeability tensor and the preset safety threshold. The calculation of the equivalent permeability tensor of the fluid layer includes: Extract the principal stress directions and stress gradient values ​​from the structural stress distribution data; The permeation anisotropy principal axis is determined based on the principal stress direction; The permeability correction factor for each principal axis direction is calculated based on the stress gradient value. Multiply the reference permeability of the fluid layer by the permeability correction factor in the corresponding direction to obtain the component values ​​of the equivalent permeability tensor; The initial grouting strategy set includes a grouting pressure adjustment scheme, a grouting material ratio scheme, and a grouting path planning scheme. Receive external input signals, determine the correction operation type based on the external input signals, and dynamically correct the initial grouting strategy set based on the correction operation type; The initial grouting strategy set is dynamically modified, and the numerical map model is updated based on the execution feedback data.

2. The numerical simulation method for grouting reinforcement and repair of building foundations according to claim 1, characterized in that, The process of obtaining the real-time state parameters of the material layer includes: Collect in-situ soil sample data of the material layer; The moisture content, void ratio, and mineral composition percentage of the in-situ soil sample data were determined. Acquire real-time environmental monitoring data of the material layer, including groundwater level fluctuations and soil temperature gradient changes. By integrating the moisture content, void ratio, mineral composition ratio, groundwater level fluctuation value, and soil temperature gradient change value, the real-time state parameters of the material layer are generated.

3. The numerical simulation method for grouting reinforcement and repair of building foundations according to claim 1, characterized in that, The calculation of the material state feature vector of the material layer includes: The real-time state parameters of the material layer are input into a pre-trained material property mapping model; The material property mapping model maps the real-time state parameters of the material layer to a vector space of fixed dimension. The output mapped vector is used as the material state feature vector, wherein the material state feature vector characterizes the comprehensive physical and mechanical properties of the material layer.

4. The numerical simulation method for grouting reinforcement and repair of building foundations according to claim 1, characterized in that, The dynamic modification of the initial grouting strategy set based on the modification operation type includes: If the correction operation type is material property correction, then the material state feature vector is recalculated and the grouting material ratio scheme is updated; If the correction operation type is structural constraint correction, then the structural stress distribution data is re-extracted and the grouting path planning scheme is updated; If the correction operation type is fluid parameter correction, then the equivalent permeability tensor is recalculated, and the grouting pressure adjustment scheme and the grouting path planning scheme are corrected simultaneously.

5. The numerical simulation method for grouting reinforcement and repair of building foundations according to claim 1, characterized in that, The step of updating the numerical graph model based on execution feedback data includes: Collect real-time monitoring data during the grouting construction process. The real-time monitoring data includes the actual diffusion radius of the grout, the grouting pressure fluctuation value, and the formation uplift. The fluid permeability coefficient of the fluid layer is updated based on the deviation between the actual diffusion radius of the slurry and the simulated diffusion path. The structural stress state of the structural layer is updated based on the degree of deviation between the grouting pressure fluctuation value and the preset pressure threshold. The material parameter variation values ​​of the material layer are inverted based on the formation uplift. Based on the updated fluid permeability coefficient, structural stress state, and material parameter changes, the dynamic attribute set of the relevant analysis nodes in the numerical graph model is recalculated.

6. The numerical simulation method for grouting reinforcement and repair of building foundations according to claim 5, characterized in that, The process of retrieving the material parameter variation values ​​of the material layer based on the formation uplift includes: Establish the inverse mapping relationship between the formation uplift and the soil compression modulus; Based on the inversion mapping relationship and the measured formation uplift, the updated value of the compressive modulus of the material layer is calculated; Based on the updated value of the compression modulus, the porosity and moisture content parameters of the material layer are corrected in reverse.

Images