A method and system for dynamically feeding back and optimizing support parameters of a sand tunnel rotary jet grouting pile

Abstract

This invention proposes a dynamic feedback optimization method and system for jet grouting support parameters in aeolian sand tunnels, relating to the fields of tunnel engineering and underground structure reinforcement technology. Based on drilling and geophysical data, a three-dimensional a priori constraint model of the aeolian sand strata is established. The section to be reinforced is divided into several construction units, and optical fibers are deployed to construct a three-dimensional optical fiber monitoring grid. Based on the construction units, jet grouting operations are initiated, using perceptible jet grout for grouting. Multi-source monitoring data and equipment parameters are collected simultaneously during the construction process, and grout characteristics are extracted by combining the optical fiber signals from the three-dimensional monitoring grid. Based on these characteristics and the aforementioned a priori constraint model, a three-dimensional pile formation effect field is reconstructed using an inversion algorithm. The inversion results are spatially compared with the design objectives, and parameters are back-calculated for insufficient areas. Model predictive control or multi-objective optimization algorithms are used to adaptively adjust the grouting parameters. This achieves visualization of the jet grouting reinforcement process, precise control, and intelligent quality control.

Description

Technical Field

[0001] This invention relates to the field of tunnel engineering and underground structure reinforcement technology, and in particular to a dynamic feedback optimization method and system for jet grouting pile support parameters in aeolian sand tunnels. Background Technology

[0002] Aeolian sandstone strata are widely distributed in arid and semi-arid regions. Their unique engineering geological characteristics—poor particle size distribution, weak cementation, high porosity, high permeability, and significant spatial fluctuations in mechanical properties—determine their extremely poor engineering stability. During tunnel and underground structure construction traversing these highly spatially variable strata, frequent disasters such as face instability, over-excavation collapse, sand inrush, and groundwater surges seriously threaten construction safety and progress. Therefore, the pre-consolidation and sealing of the surrounding rock ahead of the tunnel face is particularly crucial.

[0003] Currently, jet grouting is the mainstream solution for addressing this problem. It forms a solidified body by injecting high-pressure grout, effectively improving the overall integrity, deformation resistance, and impermeability of the formation, thus ensuring construction safety. However, in practical applications in highly variable aeolian sand formations, existing technologies still face insufficient adaptability and struggle to meet the demands of refined construction.

[0004] The shortcomings of existing technologies are as follows: the reinforcement effect is a black box, the grout diffusion path is complex and irregular, and key parameters such as pile shape and strength cannot be obtained in real time in a non-invasive manner. Existing detection methods can only provide indirect information and cannot intuitively reflect the grout diffusion boundary and solidification process. At the same time, parameter control lacks quantitative basis, and construction parameters mostly rely on experience or fixed preset values. A dynamic decision-making system based on stratum characteristics and pile feedback has not been established, resulting in high uncertainty of reinforcement effect. In addition, the response to stratum variability is lagging, and it is impossible to identify the spatial fluctuations of stratum strength and permeability in real time. It is difficult to modify the reinforcement strategy in a targeted manner, resulting in poor reinforcement uniformity, failure to effectively eliminate local risks, and the post-pile detection mode cannot achieve real-time quality control during the construction process. Summary of the Invention

[0005] To address the aforementioned issues, this invention proposes a dynamic feedback optimization method and system for jet grouting pile support parameters in aeolian sand tunnels. This transforms traditional jet grouting pile construction from an experience-dependent "black box" operation into a visualized, adaptive, and closed-loop intelligent control process based on fiber optic sensing and real-time three-dimensional inversion.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a method for dynamic feedback optimization of jet grouting support parameters in aeolian sand tunnels, comprising: Based on drilling and geophysical data, a three-dimensional a priori constraint model of aeolian sand formations was established. The section to be reinforced was divided into several construction units and optical fibers were laid out to construct a three-dimensional optical fiber monitoring grid. Based on the construction unit, the jet grouting pile operation is started, and the jet grouting slurry that can be sensed is used for grouting. Multi-source monitoring data and equipment parameters of the construction process are collected simultaneously, and the slurry characteristics are extracted by combining the three-dimensional monitoring grid fiber optic signal. Based on the aforementioned features and the prior constraint model, the three-dimensional pile formation effect field is reconstructed using an inversion algorithm. The inversion results are spatially compared with the design objectives, and the parameters of the insufficient areas are back-calculated. The grouting parameters are adaptively adjusted by model predictive control or multi-objective optimization algorithms.

[0007] Secondly, the present invention provides a dynamic feedback optimization system for jet grouting support parameters in aeolian sand tunnels, comprising: The prior constraint acquisition module is used to establish a three-dimensional prior constraint model of aeolian sand based on drilling and geophysical data. The fiber optic deployment module is used to divide the section to be reinforced into several construction units and deploy optical fibers to build a three-dimensional fiber optic monitoring grid. The feature extraction module is used to start jet grouting pile operation based on the construction unit, use perceptible jet grout for grouting, simultaneously collect multi-source monitoring data and equipment parameters during the construction process, and extract grout features by combining the three-dimensional monitoring grid fiber optic signal. The inversion module is used to reconstruct the three-dimensional pile formation effect field based on the features and the prior constraint model through an inversion algorithm. The dynamic adjustment module is used to spatially compare the inversion results with the design target, back-calculate parameters for insufficient areas, and adaptively adjust the grouting parameters using model predictive control or multi-objective optimization algorithms.

[0008] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the dynamic feedback optimization method for jet grouting support parameters in aeolian sand tunnels as described in the first aspect.

[0009] Fourthly, the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps in the dynamic feedback optimization method for jet grouting pile support parameters in aeolian sand tunnels as described in the first aspect.

[0010] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention achieves real-time, three-dimensional, and quantitative perception and control of the grouting pile formation effect by integrating distributed fiber optic three-dimensional monitoring, grout tracer, and multi-field coupled inversion. First, the grout diffusion and solidification process is acquired in real time via fiber optic signals, achieving construction transparency. Second, based on Bayesian inversion and prior models, the spatial morphology and strength field of the reinforced body can be reconstructed with high accuracy even with incomplete data, greatly improving the reliability and predictability of quality assessment. Finally, a rapid closed loop of monitoring-decision-execution is formed through model predictive control, which can automatically identify weak areas and adaptively adjust parameters, ensuring that the reinforcement quality meets standards while optimizing the grouting volume and saving costs. Overall, this invention improves the reliability, economy, and intelligence of construction in highly variable strata such as aeolian sand, and possesses the ability to continuously self-optimize based on historical data.

[0011] Advantages of additional aspects of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0012] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute a limitation thereof.

[0013] Figure 1 This is a flowchart illustrating the main process of a dynamic feedback optimization method for jet grouting pile support parameters in aeolian sand tunnel, as provided in an embodiment of the present invention. Figure 2 A detailed flowchart of a dynamic feedback optimization method for jet grouting support parameters in aeolian sand tunnel provided in an embodiment of the present invention. Detailed Implementation

[0014] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0015] Example 1 like Figure 1 As shown in the figure, this embodiment discloses a dynamic feedback optimization method for jet grouting pile support parameters in aeolian sand tunnels, including the following steps: S1: Based on drilling and geophysical data, establish a three-dimensional aeolian sand strata prior constraint model; S2: Divide the section to be reinforced into several construction units and lay optical fibers to construct a three-dimensional optical fiber monitoring grid. S3: Based on the construction unit, start the jet grouting operation, use perceptible jet grout for grouting, simultaneously collect multi-source monitoring data and equipment parameters during the construction process, and extract grout characteristics by combining the three-dimensional monitoring grid fiber optic signal. S4: Based on the features and the prior constraint model, reconstruct the three-dimensional pile formation effect field using an inversion algorithm; S5: Spatial comparison of the inversion results with the design target, back-calculation of parameters for insufficient areas, and adaptive adjustment of grouting parameters using model predictive control or multi-objective optimization algorithms.

[0016] Next, combined Figure 2 This embodiment provides a detailed description of a dynamic feedback optimization method for jet grouting pile support parameters in aeolian sand tunnels.

[0017] (I) Modeling of Stratigraphic Spatial Variability and Prior Constraints: Before construction, based on existing drilling data, high-density electrical resistivity tomography, tunnel seismic wave advance prediction (TSP), and other geophysical results, a three-dimensional a priori constraint model of aeolian sand formations was established to quantify the highly variable parameters unique to aeolian sands, such as the porosity field. n ), permeability field ( K ), internal friction angle ( φ ) and cohesion ( c Spatial distribution of ) It should be understood that the process of establishing a three-dimensional a priori constraint model for aeolian sand formations involves integrating drilling data to calibrate the stratigraphic interfaces and physical parameters, using high-density electrical resistivity tomography and tunnel seismic wave advance prediction (TSP) geophysical results to supplement the spatial distribution characteristics of the stratigraphy, and using geostatistical methods (such as Kriging interpolation and stochastic simulation) to construct a three-dimensional grid and assign spatial variation attributes, which is achievable by those skilled in the art.

[0018] The process of constructing the prior constraint model is as follows: S101: Perform unified coordinate system processing, noise reduction, and uncertainty modeling on drilling data, high-density electrical resistivity tomography (EDT) data, TSP data, etc.

[0019] S102: Based on standardized data, extract geophysical index data such as resistivity and sound velocity, as well as measured engineering parameter data such as porosity and permeability coefficient. Establish one-to-one correspondence conversion relationships between resistivity and engineering parameters, and between sound velocity and engineering parameters through statistical regression or Bayesian methods. At the same time, calculate and label the conversion errors of each conversion relationship by combining the uncertainty index quantified in S101, forming a geophysical-engineering parameter conversion model with error quantification, so as to realize the effective conversion of geophysical data into engineering parameters.

[0020] S103: Based on the measured engineering parameter samples of borehole points and the corresponding borehole spatial coordinate information in the standardized drilling data obtained in S101, the marginal distribution (numerical distribution law of a single parameter), variogram (the relationship between the magnitude of parameter variation in space and distance) and anisotropic characteristics (the variation difference of parameters in different spatial directions) of each engineering parameter are fitted. Then, a multivariate spatial correlation structure is constructed to clarify the spatial correlation law between different engineering parameters and obtain a multivariate spatial correlation structure model of engineering parameters, which provides structural constraints for subsequent spatial distribution modeling.

[0021] S104: The "soft constraint" engineering parameter field obtained by transforming the standardized geophysical data obtained in S101 through the conversion model in S102, and the "hard constraint" data formed by the measured data of the standardized drilling engineering parameters obtained in S101, are input together with the multivariate spatial correlation structure model obtained in S103 into the conditional random field simulation or Bayesian update algorithm. Through the organic integration of multi-source constraints, a three-dimensional a priori field of the formation (including the three-dimensional spatial distribution of key parameters such as porosity field and permeability coefficient field) that can truly reflect the high variability of aeolian sand is generated. Moreover, the a priori field has fully incorporated the constraint information and error characteristics of the multi-source data.

[0022] S105: Taking the complete three-dimensional prior field obtained in S104 as input, it is discretized using low-rank representation methods such as KL expansion, transforming the continuous spatial distribution into a quantifiable and computable parameterized model. The final output includes the prior mean. Prior covariance The computable prior variability model of the discretized parameter matrix (prior constraint model) is directly used as the prior constraint for the subsequent fiber optic monitoring inversion in the pile-forming stage, ensuring that the entire modeling process forms a logical closed loop.

[0023] Using the prior variability model as a Bayesian prior constraint or Tikhonov regularization term for the pile effect field inversion ensures that the inversion results can still utilize prior knowledge of the stratigraphy even when fiber optic observation data is incomplete or contains noise, thereby improving robustness and reducing uncertainty.

[0024] Among them, the prior three-dimensional variability model provides the prior means of the parameter field to be inverted. with prior covariance In the inversion process, this data is incorporated into the objective function, forming a joint optimization problem with the observed data. Its mathematical form can be expressed as:

[0025] In the formula, d represents the observed data, and m represents the parameter field to be inverted. This represents the covariance matrix of the observation data; the first term is the fitting error of the pile observations such as fiber optic / tracer, and the second term is the prior constraint, which is equivalent to the Tikhonov regularization term; when a probabilistic interpretation is adopted, this term corresponds to the Gaussian Bayes prior.

[0026] Numerically, the solution can be obtained through iterative linearization or the least squares method, or accelerated using low-rank covariance, KL decomposition, or ensemble Kalman filter. In this way, the prior model naturally serves as a constraint for the inversion, ensuring that the inversion results maintain a reasonable spatial structure while conforming to observations.

[0027] This embodiment introduces a priori three-dimensional spatial variability model of the formation (porosity, permeability coefficient, etc.) into the reinforcement control process. Traditional jet grouting parameter optimization is based on the assumption of ideal homogeneous formation, which cannot cope with the high variability of aeolian sand. This embodiment uses this variability model as a Bayesian prior constraint to achieve multi-source information fusion. Even if the fiber optic observation data has sparsity or noise, the inversion results can still utilize prior knowledge of the formation, greatly improving the robustness, accuracy, and reliability of the pile formation effect inversion, and making subsequent parameter adjustments more targeted.

[0028] (II) Construction Unit Division and Fiber Optic Deployment: Based on the tunnel axis and construction requirements (longitudinal axis and construction progress), the section to be reinforced ahead is divided into several independent construction units; for example, the unit length can be 0.5 to 2.0 meters, depending on the requirements for soil uniformity and construction progress.

[0029] Around each construction unit, a distributed three-dimensional optical fiber monitoring grid is designed and deployed along the axial, radial, and circumferential directions to cover the target reinforcement area. The location, depth, and spacing of the buried optical fibers must meet the minimum sampling density requirements of the pile formation effect field. The axial fiber spacing can be 0.5 to 1.0 meters, and the radial detection depth covers the designed reinforcement radius with a certain margin.

[0030] This embodiment constructs a distributed optical fiber three-dimensional monitoring network. By deploying high-density optical fibers in the axial, radial, and circumferential directions, it achieves continuous and three-dimensional spatial perception of the entire reinforced area. This provides a reliable spatial observation basis for the real-time three-dimensional inversion of the pile formation effect field, which helps to more accurately capture the non-uniform morphology and spatial variability of the aeolian sandstone stratum reinforced body. It overcomes the limitations of existing technologies that mainly rely on discrete points or two-dimensional plane monitoring, thereby making up for the shortcomings of traditional point monitoring or experience-based construction in terms of continuity and spatial coverage.

[0031] Furthermore, high-precision distributed optical fibers are deployed in the pre-drilled holes or pre-buried pipes in front of the tunnel face and in the surrounding rock. Based on the site requirements of aeolian sand tunnels, distributed optical fiber sensors based on Brillouin scattering or Rayleigh scattering principles are selected, which can cover a wide area and provide non-invasive three-dimensional continuous monitoring.

[0032] Before grouting begins, the entire fiber optic sensing system is calibrated to an environmental baseline. The baseline temperature, strain, background scattering intensity, and ambient acoustic spectrum along the fiber optic line are recorded to ensure that the grout signal acquired subsequently is a precise difference from the baseline signal, thereby eliminating environmental interference.

[0033] Configure high-frequency fiber optic data acquisition and demodulation equipment. The demodulation system needs to have high spatial resolution and high time sampling frequency. The sampling frequency can be set to 1~10Hz or higher to capture the instantaneous arrival time of the slurry diffusion front (i.e., real-time response).

[0034] This embodiment employs high-precision distributed optical fiber and high-frequency synchronous acquisition technology, and eliminates the interference of complex environmental factors such as temperature and strain on optical signals through environmental baseline calibration. Compared with traditional intermittent or low-frequency sensors, this embodiment can capture the instantaneous arrival time of the grout diffusion front in real time at frequencies of 1~10 Hz or even higher. Its high spatial resolution and high temporal sampling rate ensure continuous and accurate tracking of the rapid construction process of jet grouting piles, providing hardware support for achieving "dynamic" feedback.

[0035] (III) Preparation and calibration of the perceptible tracer slurry: The base material for jet grouting is prepared, including cement, fly ash admixture and thickener. The water-cement ratio (w / c) is adjusted according to the permeability characteristics and target strength of the aeolian sand strata to ensure the stability and diffusivity of the grout.

[0036] Uniformly incorporating optical fibers into a slurry allows for the sensing of chemical tracer microparticles. These microparticles possess the property of generating fluorescence or strong scattering at specific wavelengths, such as rare-earth-doped nanoparticles or functionalized fluorescent silica microspheres. The particle size and optical properties are optimized to ensure they can flow with the slurry and be recognized by the optical fiber system even at low concentrations. The mass fraction of the microparticles is controlled between 0.05% and 2%, adjusted according to the sensitivity of the optical fiber system.

[0037] In small-scale grouting tests in the laboratory or on-site, optical calibration is performed. By injecting tracer slurries of different concentrations, a quantitative mapping relationship is established between the optical response intensity of the fiber (such as fluorescence intensity and scattering attenuation) and the concentration / solidification state of the tracer slurry.

[0038] Specifically, the quantitative mapping relationship is obtained through experimental calibration: in small-scale grouting tests in the laboratory or on-site, several tracer grouts of known concentrations (or grouts at different solidification stages) are sequentially injected into the soil, while the optical response, such as fluorescence intensity and scattering attenuation, of the optical fiber is collected in real time. For each known concentration... Corresponding optical signal To fit the model, linear regression, logarithmic regression, or multinomial / exponential models can be used to obtain the desired results. or its inverse function This fitting constitutes a quantitative mapping between the optical response and the tracer slurry concentration (or solidification state), and serves as the basis for the forward model or observation operator in subsequent inversion.

[0039] This embodiment achieves non-invasive real-time visualization of the jet grout diffusion process by incorporating optical fibers into the grout to detect chemical tracer particles. In existing technologies, grout diffusion is a "black box" operation, and the pile formation effect is not visible. This embodiment, through precise optical calibration, establishes a quantitative mapping between optical fiber signals and grout concentration / solidification state, transforming the "black box" into a "gray box." This allows the optical fiber system to perceive key information such as the arrival of the diffusion front and the penetration range of the grout in real time and quantitatively, providing a direct physical signal input for dynamic feedback.

[0040] (iv) Jet grouting pile construction and tracer slurry injection: The jet grouting operation was initiated according to the divided construction units, using pre-set initial parameters, including jetting pressure. p 0. Grouting volume Q 0. Increase speed v 0. Duration of a single pile t 0. Overlap rate d 0, etc.; During the grouting process, fiber optic signals are collected in real time and synchronously, including changes in scattering / fluorescence intensity caused by tracer particles, changes in acoustic characteristics caused by grouting, changes in temperature / strain caused by curing, and grouting equipment parameters such as pressure, flow rate, and pump speed.

[0041] This embodiment achieves real-time synchronous acquisition of key construction parameters and multi-channel fiber optic sensing data. This synchronization ensures the temporal accuracy of the data, enabling the system to establish an instantaneous causal relationship between grouting parameters (such as pressure) and pile formation effects (such as diffusion signals). Unlike the post-event analysis of grouting records in existing technologies, this embodiment can analyze the impact of parameter changes on the reinforcement effect on a millisecond-level timescale, laying a data foundation for the accuracy and real-time performance of subsequent closed-loop control.

[0042] (v) Signal feature extraction and diffusion consolidation identification: The massive amount of raw fiber data is subjected to wavelet filtering for noise filtering, system error correction, and baseline correction, and temperature-strain coupling compensation is implemented to remove the influence of environmental factors. Key features for inversion are automatically extracted, including abrupt change points of the signal along the fiber line to accurately indicate the location and time of the slurry diffusion front, signal intensity attenuation curves along the distance to reflect the tracer concentration distribution, and the strain or temperature change rate of the fiber during the curing process.

[0043] Specifically, when processing massive amounts of raw fiber optic data, the process first employs Discrete Wavelet Transform (DWT) for noise filtering. A Daubechies wavelet basis is selected to decompose the signal into multiple scales, and soft thresholding is used to remove high-frequency noise while preserving key signal features. System error correction is achieved through a linear regression model based on calibration data, using known reference points to fit and correct instrument deviations. Baseline correction uses a polynomial fitting method to estimate and subtract the background trend of the signal to ensure a stable signal baseline. For temperature-strain coupling compensation, ambient temperature data is collected by deploying reference fibers and external temperature sensors. A dual-parameter decoupling algorithm (such as a frequency shift and power change model based on Brillouin scattering) is used to remove the temperature effect, thereby separating the pure strain signal.

[0044] In the feature extraction stage, abrupt change points at the arrival of the slurry diffusion front are automatically identified: a wavelet transform modulus maxima edge detection algorithm is used to scan the signal along the fiber optic line, locate points with significant amplitude changes, and record the corresponding positions and timestamps. The signal intensity attenuation curve is extracted by normalizing the filtered signal along the distance axis and fitting an exponential decay model to reflect the tracer concentration distribution. The strain or temperature change rate during the solidification process is calculated using the sliding window difference method to calculate the first derivative of the time series, capturing the time-domain characteristics caused by early consolidation. These processed key features, including abrupt change points, attenuation curves, and change rates, serve as observation inputs for subsequent inversion algorithms, supporting accurate analysis of slurry diffusion dynamics.

[0045] This embodiment can automatically and precisely extract multiple features from complex raw fiber optic data. By preprocessing and identifying the arrival point of the slurry diffusion front, the signal attenuation curve, and the consolidation time-domain characteristics, this method can not only determine whether the slurry has arrived, but also quantitatively analyze the slurry concentration distribution and the consolidation strength trend. Compared to traditional monitoring methods that typically only provide single indicators such as temperature or strain, this embodiment significantly expands the dimension and depth of observational information, providing richer data support for multi-field coupled inversion.

[0046] (vi) Real-time three-dimensional inversion of the pile formation effect field: A multi-field coupled model of tracer signal, formation variability, and consolidation mechanical response was constructed. Using the extracted fiber optic features as observations, and combining the formation prior model and the slurry material constitutive model, a Bayesian inversion or Tikhonov regularized finite element inversion algorithm was used to reconstruct the three-dimensional pile formation effect field in real time. Real-time output of slurry diffusion boundary field (equivalent reinforcement radius) R eff Alternatively, the target concentration isosurface can be used to accurately display the morphology of the solidified body and output the consolidation strength evolution field (equivalent elastic modulus) of the solidified body in real time. E eqv or compressive strength f cu Spatial distribution; By using deep learning-based networks (such as convolutional neural networks) to build surrogate models, the complex inversion process is accelerated to millisecond-level computation, enabling fast and high-precision real-time field reconstruction and real-time dynamic visualization feedback.

[0047] Specifically, the real-time three-dimensional inversion of the pile formation effect field is achieved through the following steps: First, based on the characteristics obtained from the aforementioned fiber optic signal preprocessing (leader arrival time, signal attenuation curve along the distance, and temperature / strain change rate during the curing stage), these are used as observations. .

[0048] Secondly, the prior mean of the parameter field is obtained based on the prior three-dimensional formation variability model constructed from multi-source geophysical exploration and drilling. Covariance In addition, soil parameters such as equivalent permeability and initial porosity are used; material parameters of the slurry and solidified body (such as viscosity, diffusion coefficient, and curing rate constant) are obtained by indoor tests and field small-scale tests.

[0049] Then, a forward relation is established based on Darcy's seepage, tracer transport (convection-dispersion), and the constitutive model of solidification mechanics. The front analysis yields observables of the optical fiber, such as the front position, concentration distribution, and temperature / strain response over time.

[0050] The inversion is performed using Bayesian MAP or Tikhonov regularized finite element method to solve the objective function: ; Iterative linearization (Gauss-Newton or LM), sparse PCG, and KL decomposition or a reduced-order model are used to solve the problem, and the dimensionality is reduced. Since the parameter field *m* to be inverted already contains the core parameters controlling slurry diffusion and solidification reaction, a forward model is also used. The corresponding seepage and solidification mechanical properties can be directly calculated from the m-mapping obtained by the inversion. Each step of the inversion outputs a concentration isosurface or diffusion radius. and curing strength (equivalent modulus) Compressive strength The three-dimensional distribution of ).

[0051] To achieve real-time performance, a large number of samples were obtained. For training convolutional neural network surrogate models, to approximate mapping During real-time operation on-site, a proxy model replaces part of the finite element solution, reducing the calculation time to the millisecond level, and completing the real-time reconstruction of the three-dimensional effect field and dynamic visualization output.

[0052] This embodiment constructs a multi-field coupled inversion model that integrates tracer signals, formation response, and consolidation mechanics. Combined with prior information, it achieves real-time three-dimensional reconstruction of the pile formation effect field. Compared to traditional, delayed, indirect, and primarily two-dimensional detection methods, this embodiment can output the spatial distribution of grout diffusion boundary Reff and consolidation strength (Eeqv, fcu) in real time. By introducing deep learning-based surrogate modeling technology, the complex calculation process is accelerated to the millisecond level, significantly improving real-time performance while ensuring high accuracy of the inversion results. This provides reliable support for the visual monitoring of reinforcement effects.

[0053] (vii) Identification and evaluation of spatial differences in reinforcement quality: The three-dimensional pile formation effect field obtained from real-time inversion is compared point-by-point in space with the design target field (target reinforcement radius, target strength) to calculate local performance indicators. R i → R i ( x , y , z = Actual strength / Target strength.

[0054] According to performance indicators R i The value automatically divides the monitoring area into compliance zones. R i ≥1), Area with insufficient potential (0.7≤ R i <1) Vulnerable danger zone ( R i <0.7) or excessive slurry area, to achieve refined spatial management of reinforcement quality.

[0055] This embodiment transforms engineering judgment into a quantitative spatial management method. By calculating the local effectiveness index Ri and comparing it point-by-point in space, it can accurately identify weak areas and areas with excessive grout. This method effectively improves the problem of insufficient reinforcement uniformity in traditional construction, provides a basis for implementing differentiated grouting, and thus avoids insufficient local reinforcement while reducing the excessive use of materials in strong areas.

[0056] (viii) Adaptive decision-making and back-calculation of support parameters: For identified areas with insufficient potential or weak and dangerous areas, the system uses a parameter back-calculation module to calculate the required correction parameter increment to achieve the target. , , , The inverse calculation follows the finite step size rule. ≤0.1 p 0, ≤0.2 Q 0), to prevent system overshoot.

[0057] Model predictive control (MPC) or multi-objective optimization algorithms are used for real-time decision-making. The control objective is to achieve a balance between minimizing the grouting volume and maximizing the reinforcement compliance rate. At the same time, the permeability of the formation, the stiffness of the grouting body, and local spatial differences must be considered. In weak or high-permeability areas, a reinforcement strategy of high pressure, low speed, and high overlap rate is preferred. In areas with excessive grout, parameters are automatically optimized (such as reducing pressure or grouting volume) to save materials and costs.

[0058] This embodiment introduces model predictive control or multi-objective optimization algorithms, enabling the system to adaptively adjust support parameters based on real-time acquired pile formation effect field data. Control objectives include not only achieving technical targets but also minimizing grouting volume (economic efficiency) and maximizing reinforcement compliance rate (safety). During parameter adjustment, the system follows a spatial adaptive principle, employing high-pressure reinforcement strategies in weak areas, effectively changing the traditional construction model that relies on experience or fixed parameters, and improving the scientific rigor and precision of the construction process.

[0059] (ix) Construction control execution and closed-loop iteration: The corrected parameters are sent to the jet grouting construction control system via the wireless communication module to execute the next unit or reinforcement grouting operation, completing a full dynamic feedback closed loop. This process of continuous monitoring, inversion, and decision-making is executed until all target points (especially weak and dangerous areas) in the current construction section are reached. R i All values ​​met the design requirements, achieving adaptive and precise control of the reinforcement quality.

[0060] This embodiment constructs a complete dynamic feedback closed-loop control system. By wirelessly transmitting the corrected parameters to the execution unit to drive the next round of operation, the system realizes a rapid iterative process of "real-time monitoring - intelligent decision-making - precise execution," effectively ensuring the dynamic controllability of the reinforcement process. This allows the jet grouting pile construction to adjust its operation strategy in a timely manner when geological conditions change, improving the reliability of the reinforcement effect and its adaptability to sudden geological conditions, and achieving adaptive and precise control of the reinforcement quality.

[0061] (x) Data management, 3D visualization, and learning model construction: All original signals (data from different mileages during tunnel excavation), pile formation effect field inversion results, geological models and parameter adjustment records are correlated by timestamp and spatial location and stored in a structured construction database to establish a complete and traceable construction archive.

[0062] Construct a 3D Geographic Information System (GIS) or Engineering Information Model (EIM) interface that displays the 3D distribution of the pile formation effect field, geological variability, early warning information of weak areas, and parameter optimization suggestions in real time. Provide a human-computer interaction interface to support operators in fine-tuning or intervening in parameters, thereby improving the transparency of the project.

[0063] Based on accumulated historical construction data, an incremental deep learning network is trained to iteratively learn a model. This model is used to continuously optimize the decision-making efficiency of the future parameter back-calculation module and the prediction accuracy of the pile formation effect back-calculation model, enabling the system to have the ability to accumulate experience and continuously self-optimize.

[0064] This embodiment constructs an intelligent construction management system with data traceability and 3D visualization capabilities, including a structured construction database and a real-time 3D monitoring interface. An iterative learning mechanism based on accumulated historical data training is introduced, employing adaptive algorithms such as incremental deep learning networks to achieve continuous accumulation of construction experience and autonomous optimization of the decision-making model. With the continuous accumulation of construction data, the system can dynamically improve the prediction accuracy and decision-making efficiency of the parameter back-calculation module, gradually enhancing the intelligence level of the support system in practical applications and overcoming the limitations of existing automated systems that rely on fixed rules and lack continuous learning capabilities.

[0065] (xi) Quality threshold setting and emergency intervention mechanism: The system sets upper limits for safety and quality, such as maximum allowable grouting pressure, maximum allowable displacement rate, and upper limit for the number of consecutive failures to meet the standards. The system monitors and compares these limits in real time. When a dangerous alarm is detected, such as abnormal signal or sudden large displacement of the arch, the system automatically triggers pre-set emergency measures, such as stopping grouting and reducing pressure, and issues a graded alarm to prompt human experts for review. In complex and abnormal situations, the system supports human experts to intervene and modify the control strategy. Every decision made by the expert will be recorded in the system and used as high-quality constraint data for the optimization and correction of the subsequent iterative learning model.

[0066] This embodiment constructs a hierarchical alarm and human-machine collaborative emergency control mechanism. By setting strict safety and quality thresholds and conducting real-time monitoring, the system can automatically trigger pre-set emergency response measures before a risk occurs. Simultaneously, the system retains the right to human expert intervention and feeds expert decision records as high-quality constraint data into the model learning process, forming a closed-loop optimization. By organically integrating the real-time response capability of machines with the experience-based judgment of experts, a highly reliable and trustworthy intelligent control system is constructed, improving construction safety and risk response capabilities under complex geological conditions.

[0067] This specific embodiment constructs a closed-loop control system for jet grouting pile construction with high-level intelligent features by using technologies such as coupled fiber optic sensing, chemical tracing, multi-source information fusion inversion, and spatial adaptive control. It effectively solves the long-standing technical problems of invisible effects, parameter dependence on experience, and difficulty in quality assurance in the reinforcement of aeolian sandy strata, and realizes the transformation of the construction process from experience-based judgment to data-driven.

[0068] Example 2 This embodiment provides a dynamic feedback optimization system for jet grouting support parameters in aeolian sand tunnels, including: The prior constraint acquisition module is used to establish a three-dimensional prior constraint model of aeolian sand based on drilling and geophysical data. The fiber optic deployment module is used to divide the section to be reinforced into several construction units and deploy optical fibers to build a three-dimensional fiber optic monitoring grid. The feature extraction module is used to start jet grouting pile operation based on the construction unit, use perceptible jet grout for grouting, simultaneously collect multi-source monitoring data and equipment parameters during the construction process, and extract grout features by combining the three-dimensional monitoring grid fiber optic signal. The inversion module is used to reconstruct the three-dimensional pile formation effect field based on the features and the prior constraint model through an inversion algorithm. The dynamic adjustment module is used to spatially compare the inversion results with the design target, back-calculate parameters for insufficient areas, and adaptively adjust the grouting parameters using model predictive control or multi-objective optimization algorithms.

[0069] Example 3 This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in the dynamic feedback optimization method for jet grouting pile support parameters in aeolian sand tunnel as described in Embodiment 1 above.

[0070] Example 4 This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the dynamic feedback optimization method for jet grouting pile support parameters in aeolian sand tunnel as described in Embodiment 1 above.

[0071] The steps or modules involved in Embodiments 2 to 4 above correspond to those in Embodiment 1. For specific implementation details, please refer to the relevant description section of Embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood as including any medium capable of storing, encoding, or carrying an instruction set for execution by a processor and enabling the processor to perform any of the methods in this invention.

[0072] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for dynamic feedback optimization of support parameters of a tunnel rotary jet grouting pile in a wind-blown sand, characterized in that, include: Based on drilling and geophysical data, a three-dimensional a priori constraint model of aeolian sand formations was established. The section to be reinforced was divided into several construction units and optical fibers were laid out to construct a three-dimensional optical fiber monitoring grid. Based on the construction unit, the jet grouting pile operation is started, and the jet grouting slurry that can be sensed is used for grouting. Multi-source monitoring data and equipment parameters of the construction process are collected simultaneously, and the slurry characteristics are extracted by combining the three-dimensional monitoring grid fiber optic signal. Based on the aforementioned features and the prior constraint model, the three-dimensional pile formation effect field is reconstructed using an inversion algorithm. The inversion results are spatially compared with the design objectives, and the parameters of the insufficient areas are back-calculated. The grouting parameters are adaptively adjusted by model predictive control or multi-objective optimization algorithms.

2. The method for dynamic feedback optimization of jet grouting pile support parameters in aeolian sand tunnels as described in claim 1, characterized in that, The aforementioned establishment of a three-dimensional aeolian sand strata prior constraint model based on drilling and geophysical data involves: integrating measured engineering parameters obtained from drilling with geophysical data, constructing a three-dimensional parameter field model characterizing the spatial distribution of porosity, permeability coefficient, internal friction angle, and cohesion through data fusion and spatial interpolation methods, and quantifying it into a mathematical model containing prior mean and covariance as a prior constraint for subsequent inversion calculations.

3. The method for dynamic feedback optimization of jet grouting support parameters in aeolian sand tunnels as described in claim 1, characterized in that, The fiber optic three-dimensional monitoring grid is formed by distributing fiber optic sensing lines along the axial, radial, and circumferential directions of the tunnel construction unit, creating a three-dimensional monitoring network that covers the reinforced area.

4. The method for dynamic feedback optimization of jet grouting pile support parameters in aeolian sand tunnels as described in claim 1, characterized in that, The perceptible rotary slurry is a slurry doped with optical fiber-identifiable tracer particles; The multi-source monitoring data includes optical response signals induced by the tracer particles, acoustic signals generated during the grouting process, and temperature and strain signals generated during the grout solidification process.

5. The method for dynamic feedback optimization of jet grouting pile support parameters in aeolian sand tunnels as described in claim 1, characterized in that, The slurry characteristics include: abrupt change point characteristics reflecting the arrival location and time of the slurry diffusion front, intensity curve characteristics characterizing the attenuation law of slurry concentration along spatial distribution, and strain or temperature change rate characteristics describing the curing process.

6. The method for dynamic feedback optimization of jet grouting support parameters in aeolian sand tunnels as described in claim 1, characterized in that, The process of reconstructing the three-dimensional pile formation effect field based on the features and the prior constraint model using an inversion algorithm specifically includes: Using the extracted slurry features as observation data, and combining the prior constraint model with the slurry solidification constitutive relationship, a multi-field coupled inversion model is constructed. Using Bayesian or regularized inversion algorithms, the three-dimensional spatial field of slurry diffusion range and consolidation strength distribution is reconstructed in real time.

7. The method for dynamic feedback optimization of jet grouting pile support parameters in aeolian sand tunnels as described in claim 1, characterized in that, The process of spatially comparing the inversion results with the design target, back-calculating parameters for insufficient areas, and adaptively adjusting grouting parameters using model predictive control or multi-objective optimization algorithms specifically includes: The ratio of the actual intensity to the target intensity at each point in the inversion effect field is calculated as a local effectiveness index. Based on this index, the region is divided into a qualified area, a region with insufficient potential, and a weak and dangerous area. For areas with insufficient potential and weak and dangerous areas, the parameter adjustment amounts of grouting pressure, flow rate or lifting speed are calculated based on model predictive control or multi-objective optimization algorithms, and the execution parameters of subsequent construction units are adaptively corrected.

8. A dynamic feedback optimization system for jet grouting pile support parameters in aeolian sand tunnels, characterized in that, include: The prior constraint acquisition module is used to establish a three-dimensional prior constraint model of aeolian sand based on drilling and geophysical data. The fiber optic deployment module is used to divide the section to be reinforced into several construction units and deploy optical fibers to build a three-dimensional fiber optic monitoring grid. The feature extraction module is used to start jet grouting pile operation based on the construction unit, use perceptible jet grout for grouting, simultaneously collect multi-source monitoring data and equipment parameters during the construction process, and extract grout features by combining the three-dimensional monitoring grid fiber optic signal. The inversion module is used to reconstruct the three-dimensional pile formation effect field based on the features and the prior constraint model through an inversion algorithm. The dynamic adjustment module is used to spatially compare the inversion results with the design target, back-calculate parameters for insufficient areas, and adaptively adjust the grouting parameters using model predictive control or multi-objective optimization algorithms.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps in the dynamic feedback optimization method for jet grouting pile support parameters in aeolian sand tunnel as described in any one of claims 1-7.

10. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps in the dynamic feedback optimization method for jet grouting support parameters in aeolian sand tunnel as described in any one of claims 1-7.

Images