Surface subsidence control method and system based on properties of multiple-source slag

Abstract

The application discloses a kind of ground subsidence control methods based on multi-source nature of slag, comprising: obtaining the strain distribution of initial grouting material under ground subsidence area to be measured;Based on the strain distribution, determine the target connection relationship matrix for describing strain anomaly;If the number of target connection relationship matrix is 0, end the step;The target connection relationship matrix is input into the trained graph neural network to determine the corresponding secondary grouting material;The secondary grouting material is used as the initial grouting material again.The application uses FBG sensor to monitor and accurately describe the internal stress distribution of the ground;Further introduce graph neural network to optimize the selection of secondary grouting material, and solve the problem of loss function in the training process and verification difficulty.

Description

Technical Field

[0001] This invention belongs to the field of surface settlement control, and more specifically, relates to a method and system for surface settlement control based on the properties of multi-source slag. Background Technology

[0002] In existing technologies, grouting materials often employ cement-fly ash-water glass systems or cement-bentonite systems. Their advantages include controllable setting speed, high compressive strength, and good fluidity, enabling them to quickly fill the shield tail voids and seal ground cavities during tunnel boring, preventing surface subsidence and soil intrusion. However, traditional grouting materials suffer from high costs, significant environmental impact, and are prone to segregation and backflow during construction.

[0003] To improve the resource utilization and environmental friendliness of grouting materials, a novel green grouting scheme has been proposed in existing technologies. This scheme comprehensively considers the basic working and durability properties of the grout (such as impermeability, segregation resistance, and water stability in groundwater environments), partially or completely replacing the natural sand and gravel components in the grout with excavated soil (derived from pile foundation waste, tunnel boring machine excavation, and other solid waste construction debris). More specifically, researchers have introduced appropriate admixtures (such as retarders, dispersants, and thickeners) into the cement-soil system, and through scientific experimental design, have constructed an environmentally friendly synchronous grouting material based on a cement-soil-admixture system.

[0004] Based on this, the system reveals the influence relationship between key design parameters such as water-cement ratio, waste sand ratio, and powder-ash ratio and the workability (such as fluidity, bleeding rate, and setting time) and durability of grout. Combined with the GP model (Goal Programming), the nonlinear programming tool in MATLAB is used to solve for the grouting material ratio scheme that balances optimal performance and cost, thus achieving high-performance and low-cost synergistic optimization of grouting materials.

[0005] However, although the above technologies have enabled the resource utilization of construction solid waste, the use of slag as raw material has problems such as uneven composition, large fluctuations in particle size distribution, and uncontrollable organic impurity content. This poses a great challenge to the grouting stability and filling uniformity during shield tunneling, and thus introduces new uncertainties to the control of surface settlement.

[0006] Traditional surface settlement monitoring and prediction techniques typically employ the method of pre-installing fixed monitoring points. By comparing the elevation changes of these monitoring points before and after tunnel boring machine (TBM) excavation, the settlement trend of the entire surface can be inferred. This method implicitly assumes that the spatial distribution of the ground response is continuous and derivable, neglecting the existence of spatially non-uniform settlement, and in particular, failing to consider the influence of settlement gradients.

[0007] Settlement gradient, or the rate of change of settlement per unit length, is a direct indicator of the non-uniformity of surface deformation. Its impact on structural safety far exceeds that of the average settlement, especially in areas with a high density of adjacent buildings, bridges, underground utility tunnels, shallow foundations, or old structures. Abrupt settlement gradients are often the root cause of engineering defects such as structural cracking, pipeline breakage, and foundation slippage.

[0008] Under conditions of high slag content, due to poor uniformity of grouting materials and large fluctuations in filling density, the stratum response loses its regularity. Traditional monitoring point interpolation or surface fitting methods cannot effectively identify local high gradient areas or settlement anomalies, resulting in a significant decrease in prediction accuracy and a significant increase in engineering risk. Summary of the Invention

[0009] To address the shortcomings of existing technologies, the present invention aims to overcome the aforementioned deficiencies and propose a surface settlement control method and system based on the properties of multi-source slag.

[0010] The present invention adopts the following technical solution.

[0011] The first aspect of this invention discloses a method for controlling surface settlement based on the properties of multi-source spoil, comprising:

[0012] To obtain the strain distribution in the area to be tested for surface settlement under the initial grouting material;

[0013] Based on the strain distribution, determine the target connectivity matrix used to describe the strain anomaly; if the number of target connectivity matrices is 0, then the step ends.

[0014] The target connection matrix is ​​input into the trained graph neural network to determine the corresponding secondary grouting material; the secondary grouting material is then used as the initial grouting material again.

[0015] Furthermore, the surface settlement control method based on the properties of multi-source waste soil is characterized in that the method includes:

[0016] To obtain the strain distribution in the area to be tested for surface settlement under the initial grouting material;

[0017] Based on the strain distribution, determine the target connectivity matrix used to describe the strain anomaly; if the number of target connectivity matrices is 0, then the step ends.

[0018] The target connection matrix is ​​input into the trained graph neural network to determine the corresponding secondary grouting material; the secondary grouting material is then used as the initial grouting material again.

[0019] Furthermore, the strain distribution is the center wavelength vector measured by the FBG sensor.

[0020] Furthermore, the measurement of the center wavelength vector specifically includes:

[0021] Determine the heights of multiple targets in the area to be measured;

[0022] At each target height, a corresponding FBG sensor line is deployed, wherein each FBG sensor line is an optical fiber with multiple settings of multiple periodic refractive index modulation regions pre-written.

[0023] The center wavelength of the reflected light at each target height is obtained sequentially and integrated into multiple center wavelength vectors as the strain distribution.

[0024] Furthermore, the target height includes at least a first target height and a second target height, which are located at the boundary between the multi-source slag grouting site and the original ground surface. The first target height is set within the multi-source slag grouting area, and the second target height is set within the original ground surface.

[0025] Furthermore, the target connection matrix is ​​determined based on the center wavelength vector; the process specifically includes: when the value of a certain element in the center wavelength vector exceeds the preset threshold range, the initial vector of the strain direction corresponding to that element is used as the boundary, and the adjacent nodes are uniformly included in the target connection matrix.

[0026] Furthermore, the target connectivity matrix is ​​used to determine the injection location of the secondary grouting material.

[0027] Furthermore, the graph neural network takes the strain distribution of the current iteration and the strain distribution of the grouting material as input, and the selection probability of the grouting material as output. The loss function is determined by the target control value in the target connection matrix of the next iteration and the probability of selecting the secondary grouting material.

[0028] Furthermore, the amount of secondary grouting material used is determined based on the selection probability of the secondary grouting material.

[0029] A second aspect of the present invention discloses a terminal, including a processor and a storage medium; characterized in that:

[0030] The storage medium is used to store instructions;

[0031] The processor is configured to operate according to the instructions to perform the steps of the method described in the first aspect.

[0032] A third aspect of the present invention discloses a computer-readable storage medium having a computer program stored thereon, characterized in that the program, when executed by a processor, implements the steps of the method described in the first aspect.

[0033] The beneficial effects of the present invention are as follows: Compared with the prior art, the present invention has the following advantages:

[0034] This invention utilizes FBG sensors for real-time monitoring and precise description of stress distribution within the Earth's surface. Traditional methods often rely on discrete monitoring points or low-resolution equipment, making it difficult to capture minute changes in localized stress concentrations. FBG sensors, however, offer advantages such as high sensitivity, high stability, and resistance to electromagnetic interference. They can be continuously deployed along optical fibers to collect data in real-time at multiple key target heights (such as the boundary between multi-source grouting areas and the original surface). By normalizing the center wavelength vector output by the FBG sensor, the distribution characteristics of the complex stress field within the Earth's surface are accurately reflected, improving the reliability of the monitoring data. Based on the precise description of surface stress distribution using FBG sensors,

[0035] In graph neural networks, the initial vector of the strain direction of the grouting material itself is coupled with the initial vector of the strain direction of the current node, achieving precise alignment between material properties and real-time strain conditions. This not only improves the graph neural network's ability to predict the response of various materials in different strata and settlement environments but also provides richer criteria for material selection at each node, thereby improving the overall system's prediction accuracy. By incorporating the target control value in the next iteration into the loss function design, the graph neural network model achieves real-time feedback correction of each iteration's output. This design allows the loss function to more accurately reflect the difference between the current selection and the actual future control effect, alleviating the difficulties of traditional single-variable loss functions in verifying multiple spatial indicators. Furthermore, each graph neural network focuses on the strain data and settlement characteristics of its corresponding level (first target height), making the prediction of grouting level distribution more targeted and accurate. Based on the target connectivity matrix, a customized loss function can be used to optimize the unique performance of each layer in stress concentration and settlement anomalies, enabling differentiated configuration of grouting materials and construction parameters for different soil layers, thus meeting the specific needs of reinforcement and consolidation at each level. Attached Figure Description

[0036] Figure 1 This is a flowchart of a surface settlement control method based on the properties of multi-source slag soil according to an embodiment of the present invention;

[0037] Figure 2 This is a schematic diagram of the internal structure of the FBG sensor according to an embodiment of the present invention. Detailed Implementation

[0038] The present application will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention, and should not be construed as limiting the scope of protection of the present application.

[0039] This invention discloses a method for controlling surface settlement based on the properties of multi-source spoil, such as... Figure 1As shown, it includes steps 1 to 3.

[0040] Step 1: Obtain the strain distribution in the area to be tested for surface subsidence.

[0041] The strain distribution should describe the strain and direction of strain at multiple points in the area to be tested for surface settlement. Strain at each point: Strain describes the degree of deformation per unit length of a material (such as soil) under external load. The value of strain reflects the magnitude of local soil deformation; early detection of abnormally increased local strain can indicate potential settlement problems or structural weaknesses. The direction of strain helps engineers determine whether the deformation caused by settlement is uniform or has a distinct principal stress direction, thus distinguishing between tensile, compressive, or shear deformation. Understanding the principal strain direction provides a basis for subsequent support, grouting, or other reinforcement measures. For example, if the strain in one direction is significantly greater than in others, more concentrated reinforcement measures may be needed in that direction. In some embodiments, the strain distribution is the center wavelength vector measured by an FBG sensor, which is composed of the center wavelengths of different periodic refractive index modulation zones. Specifically, the measurement of the center wavelength vector may include steps 1.1 to 1.3.

[0042] Step 1.1: Determine the heights of multiple targets in the area to be measured.

[0043] The target height includes at least a first target height and a second target height, which are typically located at the boundary between the multi-source grouting area and the original ground surface. This area is usually a stress concentration zone, prone to localized abnormal settlement and high strain gradients. In some embodiments, the first target height can be set within the multi-source grouting area, while the second target height can be set within the original ground surface. At each target height, its height is considered as a key control variable to ensure that the influence of different heights on stress concentration and settlement characteristics is accurately reflected in subsequent analyses. Furthermore, setting the second target height below the original ground surface, which serves as a natural boundary, not only helps to identify the settlement gradient relative to the underground grouting area but also captures the most intuitive changes in stress concentration and localized abnormal settlement.

[0044] It should be noted that the number of target heights is a fixed value, which can usually be set to 2 to 4; if it is set to 3, the added third target height should be an extension of the first target height, that is, the number of first target heights can be one or more, and different first target heights are used to determine different levels of grouting.

[0045] Step 1.2: At each target height, deploy the corresponding FBG sensor line, wherein each FBG sensor line is an optical fiber with multiple pre-written periodic refractive index modulation regions.

[0046] Considering the uneven composition, large fluctuations in particle size distribution, and uncontrollable organic impurity content of multi-source slag, in step 1.2, measurements can be performed based on a high-precision fiber Bragg grating (FBG) sensor.

[0047] The working principle of FBG sensors is essentially based on the fiber Bragg grating effect, such as... Figure 2 As shown, its core is a periodic refractive index modulation region, which reflects light of a specific wavelength, namely the Bragg wavelength λ. B As shown in the following formula:

[0048] λ B =2n c A

[0049] Where, n c A is the effective refractive index in the optical fiber, and A is the grating period.

[0050] Understandably, strain causes the optical fiber to stretch or compress, thus directly altering the physical period of the grating; that is, under stretching, A increases, while under compression, A decreases. Therefore, the magnitude of the strain can be determined based on the center wavelength of the emitted light.

[0051] exist Figure 2 In the fiber cladding: an outer low-refractive-index material that ensures light transmission within the fiber core; fiber core: a uniformly conductive region for broadband optical signals; periodic refractive index modulation region: a periodic refractive index modulation region within the fiber core, indicated by short horizontal lines representing the modulation period; incident light: broadband optical signals entering the fiber from the left; reflected light: when a wavelength in the broadband light approaches the Bragg wavelength, it undergoes constructive interference of multiple modulation periods within the periodic refractive index modulation region, resulting in strong reflection, and this reflected light returns to the source; transmitted light: other light, not at the Bragg wavelength, continues to propagate.

[0052] Understandably, an FBG sensor can only measure strain changes at its location, or more accurately, at a single point in the periodic refractive index modulation region.

[0053] It is important to note that both ends of each periodic refractive index modulation zone in each optical fiber must be securely fastened to a steel frame, and the distance between the two ends should be consistent. The fixing frame uses rectangular steel structural components, and the cross-section of the fixing frame profile can be channel steel or I-beams. Each FBG sensor is fixed to the fixing frame by mechanical clamps, and protective rubber pads and weather-resistant protective covers are embedded around the sensors to ensure data stability under the conditions of shield tunneling vibration and groundwater infiltration. The strain signals collected by the fiber optic sensors are received by a fiber optic data acquisition box integrated on the side of the base. The acquisition box is connected to the central data processing equipment via fiber optic cables to achieve real-time data transmission. The central data processing equipment is usually installed in the shield tunneling control center or a nearby monitoring room. The data acquisition box adopts a waterproof and dustproof design and has a seismic-resistant reinforcement structure. The fixing frame is fixed to the concrete base by reserved mechanical support bolts and adopts an underground installation process to ensure that the sensor network is tightly integrated with the strata of the monitoring area, thereby accurately reflecting the strain and strain direction of the grouting material.

[0054] Step 2 essentially implements the principle of Wavelength Division Multiplexing (WDM). Within each periodic refractive index modulation region along the optical fiber, a center wavelength can be generated within the spectral range of a tuned or broadband light source. Crucially, the spacing between adjacent center wavelengths must be sufficiently large to ensure that the maximum spectral shifts of adjacent periodic refractive index modulation regions do not overlap under the influence of the measured parameter; otherwise, measurement results may become blurred or erroneous. Based on this, a spectral window can be defined, representing the span of center wavelengths that the FBG sensor may cover within a specific range of measured variables. Finally, by dividing the overall spectral width or span of the light source by the spectral window width of a single sensor, the maximum number of periodic refractive index modulation regions in a single optical fiber can be estimated. Understandably, the number of optical fibers is the quotient obtained by dividing the number of periodic refractive index modulation regions in the FBG sensor circuit by the maximum number of periodic refractive index modulation regions in a single optical fiber.

[0055] In the optical path design, the reflected light is first collected by an optical circulator or optical coupler placed between the light source and the periodic refractive index modulation region, and then transmitted to the photodetector. In this way, the reflection peak wavelength generated by each periodic refractive index modulation region corresponds to the peak signal on the detector. Another approach is to use a broadband light source, ensuring that the center wavelengths of all periodic refractive index modulation regions fall within the spectral range of the light source; then, the reflected light is transmitted to a spectrometer, and the reflection peak, i.e., the center wavelength, of each periodic refractive index modulation region is determined by analyzing the spectral data.

[0056] Understandably, an FBG sensor circuit can determine the spatial distribution of its internal periodic refractive index modulation zone in a horizontal two-dimensional direction. However, since the actual shape of the target area to be grouted may have uneven height, the FBG sensor circuit does not need to be kept at the same height. Instead, its vertical position can be adaptively adjusted according to the actual situation of the grouting opening to better match the on-site construction requirements.

[0057] Step 1.3: Sequentially obtain the center wavelength of the reflected light at each target height, and integrate them into multiple center wavelength vectors as the strain distribution.

[0058] The output of step 1.3 is multiple center wavelength vectors, including a first center wavelength vector and a second center wavelength vector, corresponding to the first FBG sensor line and the second FBG sensor line, respectively. Each element in each center wavelength vector corresponds to a periodic refractive index modulation region. The value of the element is the numerical value of the center wavelength. Since strain causes the optical fiber to stretch or compress, when preprocessing the center wavelength vectors (e.g., step 2.1 below), the value of the element can be the difference between the numerical value of the center wavelength and the Bragg wavelength of the periodic refractive index modulation region, and then the result is obtained by normalization. The normalization here refers to using a scaling factor to correlate unit wavelength change with unit strain change, so that the two have a one-to-one numerical conversion relationship, thereby ensuring that the wavelength change output by the fiber optic sensor can accurately reflect the actual strain change.

[0059] Step 2: Based on the strain distribution, determine the target connectivity matrix used to describe the strain anomaly; if the number of target connectivity matrices is 0, then the step ends.

[0060] In some embodiments, the target connectivity matrix C can be determined based on the center wavelength vector. Each target connectivity matrix should have a fixed size, such as 9*9 or 11*11, but generally should not be smaller than 9*9. Its size depends on the density of the periodic refractive index modulation regions in the FBG sensor circuitry. Understandably, if the number of target connectivity matrices is 0, it indicates that the current strain distribution meets the expected requirements and no control is needed.

[0061] Based on the center wavelength vector, the target connectivity matrix is ​​determined. Specifically, when the value of any element in the center wavelength vector exceeds a preset threshold range (e.g., center wavelength ± 0.8 nm), the initial strain direction vector corresponding to that element is used as the boundary, and adjacent nodes are uniformly included in the target connectivity matrix. Understandably, the threshold range can usually be centered at the Bragg wavelength, and the element value at the very center of the target connectivity matrix corresponds to the initial strain direction vector of that element.

[0062] Understandably, in a single measurement, the number of target connectivity matrices can be zero or at least one.

[0063] Step 3: Input the target connection matrix into the trained neural network model to determine the corresponding secondary grouting material; use the secondary grouting material as the initial grouting material again, and then return to step 1.

[0064] The training process of the graph neural network includes steps S21 to S26. The graph neural network is mainly used to analyze the guiding results of different secondary grouting materials, especially the strain direction.

[0065] In a graph neural network, an initial vector of strain direction represents a node in the graph neural network, denoted as a. i The j-th element of the initial vector for the i-th strain direction is the value of the i-th element of the j-th center wavelength vector. The value of each element in the target connectivity matrix is ​​a. i [w], the position of the element depends on the initial vector of the strain direction. For ease of description, the value of the central element in the target connectivity matrix is ​​taken as the target control value.

[0066] Understandably, the initial strain direction vector actually corresponds to a set of center wavelengths vertically distributed along the periodic refractive index modulation region. Therefore, in the above text, adjacent nodes refer to the other strain direction initial vectors that are closest to each other in the horizontal two-dimensional direction. It is not difficult to deduce that the positional relationship of the elements in the target connection matrix also corresponds one-to-one with the spatial distribution of the periodic refractive index modulation region in the horizontal two-dimensional direction. Unless otherwise specified, 'a' in the following text... i The initial vector of strain direction, or 'a', refers to the vector 'a' defined by the target connectivity matrix. i Alternatively, the strain direction initial vector, w represents the number of the element in the strain direction initial vector. In the embodiments of the present invention, the value range of w is limited to the range of the first target height, i.e. w∈[0,N-1), where N is the length of the strain direction initial vector.

[0067] The difficulty in training a graph neural network lies in two aspects: firstly, the goal is to find the most suitable secondary grouting material; secondly, a suitable loss function cannot be found to verify the correctness of the graph neural network. Because there is no multispace, a method with strict control over variables (the only variable being the secondary grouting material) can be used to verify the performance of each grouting material individually to validate the output probability of the graph neural network.

[0068] Understandably, surface settlement control systems include a database of various slag components to record the proportions and properties of available grouting materials. For example, several different grouting materials may include the following proportions: First grouting material: cement: 70g; slag: 20g; fly ash: 10g; water: 40g; high-efficiency water-reducing agent (polycarboxylate): 0.2g. Second grouting material: cement: 60g; slag: 30g; fly ash: 10g; water: 45g; retarder (sodium gluconate): 0.3g. Third grouting material: cement: 80g; slag: 10g; fly ash: 10g; water: 35g; anti-separation agent (hydroxypropyl methylcellulose, HPMC): 0.15g. Fourth grouting material: cement: 50g; slag: 40g; fly ash: 10g; water: 50g; waterproofing agent: 0.4g. The fifth grouting material mix proportions include: cement: 65g; slag: 25g; fly ash: 10g; water: 42g; early strength agent (calcium chloride): 0.7g.

[0069] The surface settlement control system of this invention, based on real-time strain distribution, adjusts the grouting material (i.e., secondary grouting material) and its corresponding injection position for the next round after the current iteration, according to the target connection matrix. Understandably, the position of the target control value in the target connection matrix of this iteration serves as the injection position (i.e., grouting hole) of the secondary grouting material; that is, the injection position corresponds one-to-one with the target connection matrix.

[0070] Step S21: Based on the preset feature transformation matrix, the initial vector of strain direction is mapped to a unified feature space to obtain a spatial mapping vector, as shown in the following equation:

[0071] s i =V a ·a i

[0072] Among them, V a It is a feature transformation matrix, whose elements are initialized randomly, s i It is a spatial mapping vector.

[0073] Understandably, during the training process of the graph algorithm, the initial vector of strain direction should be the initial vector of strain direction in the historical sample data.

[0074] Step S22: For each non-edge node, calculate the associated embedding representation of its neighboring nodes.

[0075]

[0076] Where, π iThis refers to the set of node indices of the 8 nearest neighbors of node i. A non-edge node is a node in the target connection matrix that has 8 nearest neighbors. C is the target connection matrix, and R... i For the association embedding representation, both t and i represent node numbers. Understandably, if the target connection matrix is ​​v*v, then the number of non-edge nodes is (v-2)*(v-2).

[0077] Step S23: Calculate the maximum value of each non-edge node and its associated embedding representation as the attention score.

[0078] Attention score g i The direction of strain is described as follows:

[0079]

[0080] Where σ(·) represents the activation function, which can be the sigmoid function, || represents the concatenation operation between vectors, v is the preset attention vector, and is the parameter to be learned, whose initial value is randomly set.

[0081] Step S24: Calculate the coupling degree of any node with respect to any grouting material.

[0082] The coupling degree is shown in the following formula:

[0083] β ik =σ(v·softmax((g) i ·s i )||(v a ·A k )))

[0084] Among them, A k It is the initial vector of the strain direction of the k-th grouting material.

[0085] Understandably, A k In practice, this is based on prior experimental data. That is, the k-th type of grouting material is injected into a steel container, and then the same technical means (i.e., the techniques in steps 1.1 to 1.3) are used to obtain the corresponding initial strain direction vector. For feature uniformity, A is typically... k The vector length is greater than a i One less, because a i The last element (i.e. the element corresponding to the second target height) is embedded in the local table.

[0086] Step S25: Calculate the coupling feature vector based on the coupling degree.

[0087]

[0088] Among them, zk Let N be the coupling feature vector, and N be the number of elements in the target connectivity matrix.

[0089] Step S26: Calculate the strain direction vector.

[0090] The strain direction matrix G can be represented by the following equation:

[0091] Q k =V Q z k

[0092] Y k =V Y z k

[0093] L k =V L z k

[0094]

[0095] Where d is V Q The dimension, V Q V Y V L This is the weight matrix in the attention mechanism model, consisting of parameters to be learned. Its initial values ​​are randomized and used to calculate the query vector Q. k , key vector Y k Sum vector L k K is the amount of grouting material, T represents the transpose of the matrix, and G is the strain direction vector.

[0096] The loss function is:

[0097] Loss=-{(C m -C′ m )ln(y′ l )+C′ m ln(1-y′ l )}

[0098] y′=[y′1...y′ k ...y′ K = softmax(MLP(G))

[0099] Where, y′ l is an element in y′, where l represents the selected number in this iteration, and y′ is the prediction result output by the graph neural network, representing the selection probability of the grouting material; MLP() is a multilayer perceptron, composed of neurons (nodes) at multiple levels, and is a type of feedforward neural network; C m C′ is the target control value of the target connectivity matrix C in this iteration.m It is the target control value of the target connectivity matrix C′, where C′ is the target connectivity matrix for the next iteration.

[0100] It's important to note that since the number of target connection matrices corresponds one-to-one with the number of first target heights, the above essentially trains multiple graph neural networks using the same method. The number of graph neural networks should correspond one-to-one with the number of first target heights; the only difference lies in the loss function. Each different graph neural network is used to determine the hierarchical distribution of the grouting.

[0101] Taking a target height of 3 as an example, the layers are divided into shallow, medium, and deep layers, with differences in injection depth, borehole diameter, grouting pressure, and grouting rate. Generally, deeper layers require higher grouting pressure to overcome the higher compressive strength and pore resistance of the deep soil, ensuring the grout can penetrate to the target area. Furthermore, the grout injection rate is usually set faster to quickly fill soil voids under high pressure, while simultaneously ensuring sufficient time for uniform grout diffusion. Shallow layers require relatively lower grouting pressure to avoid ground disturbance or localized soil damage caused by excessive pressure. A slower grouting rate is set to more precisely control the grout diffusion range and stabilize the strata, while reducing localized disturbances caused by excessive flow velocity. It is understandable that during the iteration process, the setting of parameters such as grouting pressure and grouting rate depends on the currently selected secondary grouting material, and their values ​​can be determined from the performance indicators such as fluidity, viscosity, and setting time of the secondary grouting material pre-recorded in the slag and soil database. Given that the above details are not closely related to the core inventive points of this application, they will not be elaborated further here.

[0102] In practice, steps 1-3 are essentially iterative processes. Therefore, the loss function can be evaluated using the results of the next iteration. In other words, the graph neural network takes the strain distribution of the current iteration and the strain distribution of the grouting material as input, and the selection probability of the grouting material as output. The output is determined by the target control value in the target connectivity matrix of the next iteration, and the probability of selecting a secondary grouting material (the probability of not selecting a secondary grouting material is 1-y′). l Determine the loss function.

[0103] Understandably, during the initial training phase, different grouting materials should be selected as diversely as possible to enrich the sample data for model training in the slag and soil database. In later use, the resulting data can also serve as historical data for continued model training and iterative optimization.

[0104] In some embodiments, the amount of secondary grouting material used in each iteration can be based on the selection probability of the secondary grouting material, i.e., y′. l The amount of secondary grouting material used can be determined as follows: More specifically, the amount of secondary grouting material used can be represented by the following formula:

[0105] L′=y′ l ×L(t)

[0106] Where L represents the maximum reference grouting volume, t represents the number of iterations, and the value of L(t) is set based on experience. In principle, it should be as small as possible: when set to a small value, the measurement accuracy of the FBG sensor should be taken into account; in addition, L(t) should gradually decrease as t increases, that is, the amount of secondary grouting material used should show an exponential distribution with respect to the number of iterations.

[0107] The applicant of this invention has provided a detailed description of the embodiments of the invention in conjunction with the accompanying drawings. However, those skilled in the art should understand that the above embodiments are merely preferred embodiments of the invention. The detailed description is only intended to help readers better understand the spirit of the invention and is not intended to limit the scope of protection of the invention. On the contrary, any improvements or modifications made based on the inventive spirit of the invention should fall within the scope of protection of the invention.

Claims

1. A method for controlling surface settlement based on the properties of multi-source spoil, characterized in that, The method includes: To obtain the strain distribution in the area to be tested for surface settlement under the initial grouting material; Based on the strain distribution, determine the target connectivity matrix used to describe the strain anomaly; if the number of target connectivity matrices is 0, then the step ends. The target connection matrix is ​​input into the trained graph neural network to determine the corresponding secondary grouting material; the secondary grouting material is then used as the initial grouting material again. The strain distribution is the center wavelength vector obtained from the FBG sensor; The target connection matrix is ​​determined based on the center wavelength vector; the process specifically includes: when the value of an element in the center wavelength vector exceeds the preset threshold range, the initial vector of the strain direction corresponding to that element is used as the boundary, and the adjacent nodes are uniformly included in the target connection matrix.

2. The surface settlement control method based on the properties of multi-source waste soil according to claim 1, characterized in that, The measurement of the center wavelength vector specifically includes: Determine the heights of multiple targets in the area to be measured; At each target height, a corresponding FBG sensor line is deployed, wherein each FBG sensor line is an optical fiber with multiple settings of multiple periodic refractive index modulation regions pre-written. The center wavelength of the reflected light at each target height is obtained sequentially and integrated into multiple center wavelength vectors as the strain distribution.

3. The surface settlement control method based on the properties of multi-source spoil as described in claim 2, characterized in that, The target height includes at least a first target height and a second target height, which are located at the boundary between the multi-source slag grouting site and the original ground surface. The first target height is set within the multi-source slag grouting area, and the second target height is set within the original ground surface.

4. The surface settlement control method based on the properties of multi-source spoil as described in claim 1, characterized in that, The target connectivity matrix is ​​used to determine the injection location of the secondary grouting material.

5. The surface settlement control method based on the properties of multi-source spoil as described in claim 1, characterized in that, The graph neural network takes the strain distribution of the current iteration and the strain distribution of the grouting material as input, and the selection probability of the grouting material as output. The loss function is determined by the target control value in the target connection matrix of the next iteration and the probability of selecting the secondary grouting material.

6. The surface settlement control method based on the properties of multi-source spoil as described in claim 4, characterized in that, The amount of secondary grouting material used is determined based on the selection probability of the secondary grouting material.

7. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-6.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1-6.

Images