Optimization processing method for surrounding rock displacement monitoring data during the excavation of cavern complex

Abstract

This invention relates to an optimized processing method for monitoring surrounding rock displacement during the excavation of a cavern complex. The method includes: determining the rock mass strength parameters and establishing a three-dimensional geological model by combining topographic information of the target area and on-site geological survey data; calculating the corresponding three-dimensional geostress field; determining the approximate range of surrounding rock strength and deformation parameters according to specifications and rock mechanics handbooks based on the surrounding rock category determined by surrounding rock classification and surrounding rock parameters obtained from laboratory tests; calculating the relative deformation of different measurement sections and inverting the optimized surrounding rock parameters using the relative deformation of different measurement sections monitored by multi-point displacement gauges; and verifying the reliability of the inverted surrounding rock parameters by comparing them with the measured relative deformation characteristics of the surrounding rock. The beneficial effects of this invention are: by optimizing the surrounding rock parameters to obtain the excavation response characteristics of the surrounding rock, the on-site monitoring data is corrected and analyzed to correct the multi-point displacement gauge monitoring data, making the monitoring data obtained by the multi-point displacement gauges more accurate.

Description

Technical Field

[0001] This invention belongs to the field of rock mechanics and engineering technology, specifically relating to an optimized processing method for monitoring surrounding rock displacement during the excavation of a group of caverns. Background Technology

[0002] The study of the excavation response characteristics of surrounding rock is of great significance in underground engineering, providing a theoretical basis for the design, excavation, and support optimization of underground caverns. Multi-point displacement gauges are commonly used engineering monitoring instruments to measure displacement changes at multiple locations or depths. They are widely used in civil engineering, geological engineering, and hydraulic engineering, playing a crucial role, especially in monitoring the deformation of surrounding rock, slope stability, and foundation settlement in underground caverns. Their monitoring data can be used to monitor the convergence of surrounding rock deformation, comprehensively analyze the deformation data from different monitoring sections, determine the degree and stage of deformation by referring to deformation intervals, and classify deformation; or it can be used for excavation feedback analysis to support the assessment of the stability of surrounding rock in underground caverns. Since the initial deformation cannot be monitored, in actual engineering, the deformation increments of measuring points before and after excavation during layered excavation are generally used to analyze the inversion of surrounding rock mechanical parameters, combined with the geostress field, to evaluate and predict the stability of surrounding rock during subsequent excavation. The deformation increment of the measuring point often uses the absolute deformation of that point. However, there are disagreements regarding the determination of fixed points in the process of solving for absolute deformation, thus raising questions about the reliability and accuracy of the parameter inversion results.

[0003] Therefore, in view of the drawbacks of the above monitoring data method, an optimized processing method for the surrounding rock displacement monitoring data during the excavation of the cavern group is proposed. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this application provides an optimized processing method for monitoring surrounding rock displacement during the excavation of a cavern complex, comprising the following steps:

[0005] Step S1: Based on the on-site geological survey data, determine the strength parameters of the rock mass and the coordinate data and elevation information of the target area to establish a three-dimensional geological model;

[0006] Step S2: Based on the geostress data of the observation points in the target area obtained by field measurement, the corresponding three-dimensional geostress field is calculated by multiple linear regression method;

[0007] Step S3: Obtain the range of surrounding rock strength deformation parameters. The surrounding rock strength deformation parameters include Young's modulus, Poisson's ratio, cohesion, and internal friction angle. Each surrounding rock strength deformation parameter is divided into t different factor levels with the same number. t groups of orthogonal test groups of surrounding rock strength deformation parameters are designed through orthogonal test.

[0008] Step S4: Input the three-dimensional geostress field of the target area and the surrounding rock strength deformation parameters of different groups into the numerical model, calculate the displacement response field of the surrounding rock, extract the displacement at each location of multiple field-deployed multi-point displacement gauges based on the displacement response field, and calculate the relative deformation of the measuring segments between different locations.

[0009] Step S5: Train the BP neural network using the nonlinear relationship between the surrounding rock strength deformation parameters calculated in step S4 and the corresponding relative deformation of different test sections to obtain the BP neural network model. Input the actual relative deformation of different test sections obtained by multi-point displacement gauge monitoring into the BP neural network model to invert and obtain the optimal surrounding rock strength deformation parameters.

[0010] Step S6: Input the preferred surrounding rock deformation strength parameters and the three-dimensional geostress field obtained in step S2 into the numerical model to obtain the excavation response field of the surrounding rock through calculation.

[0011] Step S7: Based on the multi-point displacement gauge monitoring data obtained on site, the monitoring data is corrected by the excavation response field of the surrounding rock.

[0012] Step S1, the modeling process, includes:

[0013] Step 11: Download the DEM elevation information of the target area from the satellite-acquired geographic information data and import it into the 3D modeling software;

[0014] Step 12: Analyze the DEM data to extract the outline of the land surface. Using the extracted outline and elevation information, combined with the terrain coordinate data, the software generates a three-dimensional mesh.

[0015] Step 13: Stretch the generated 3D mesh along the vertical direction to the corresponding elevation to form a 3D terrain entity model. Then, cut and mesh the terrain entity model to generate a 3D geological model suitable for numerical simulation analysis.

[0016] Step S2 includes:

[0017] Step 21: Construct the geostress field. The tectonic modes include: (a) self-weight, (b) X-direction compression, (c) Y-direction compression, (d) XY horizontal shear, (e) vertical shear tectonic factors on the ZY plane and vertical shear tectonic factors on the ZX plane. Calculate the stress generated by tectonic modes (a) to (e) using finite difference numerical simulation software.

[0018] Step 22: Using the principle of linear superposition under elastic working conditions, calculate the initial ground stress value:

[0019]

[0020] in: The values ​​are calculated based on the regression of j stress components from k observation points, where j∈[1,6] represents the six stress components generated by tectonic modes (a)~(e); L i (i = 1, 2, 3, ..., n) are regression coefficients, n = 6; σ xx σ yy σ xy σ xz σ yz The stresses generated by the construction mode at the k-th observation point in five directions; σ w This represents the stress generated by gravity at the kth observation point; It is the numerical calculation value of the j-th stress component at the k-th observation point under the i-th partial load factor;

[0021] Step 23: Calculate the measured values ​​of geostress and calculated value The error S between 残 :

[0022]

[0023] in, It is the measured value of the j-th stress component at the k-th observation point, where j∈[1,6];

[0024] Step 24: Obtain the observed values and calculated value The regression coefficient L obtained from the minimum error between them i (i = 1, 2, 3, ..., n), n = 6;

[0025] According to the least squares principle, S 残 The system of equations that yields the minimum value is:

[0026]

[0027] By solving this system of matrix equations, all regression coefficients can be obtained.

[0028] In step S4, the displacement of the orifice calculated by a multi-point displacement gauge is D0, and the displacement of the r-th observation point of the multi-point displacement gauge is D. r Then, its relative deformation for the 0th to rth measurement segment of the multi-point displacement gauge is D0-D. r .

[0029] Step S5, training the BP neural network, includes:

[0030] Step 51: Take the relative deformation of different sections of the multi-point displacement gauge calculated in step S4 as the input data set xt, and take the surrounding rock deformation and strength parameters of the different sets corresponding to it in step S4 as the output data yt. Use the input data and output data as training data, and obtain the nonlinear relationship f(x) between the input data and output data by training a BP neural network:

[0031]

[0032] In the formula, x represents the input node representation of the neural network; y represents the output node representation of the neural network; f(v, h1, ..., h) p (w) is the established multi-layer neural network structure, where v is the input layer F. x The number of nodes, h1 is the number of units in the intermediate layer F1, h p For the intermediate layer F p The number of units, w is the output layer F y The number of nodes, h v R is the number of units in the intermediate layer of the v-th layer; v →R w Represents R v To R w Mapping;

[0033] The root mean square error (RMSE) is selected as the error function. Based on the RMSE of the output layer, the weights and biases of the intermediate layers are adjusted in reverse until the error converges. The error function is the square root of the sum of the squared averages of the differences between all true and predicted values. The calculation formula is as follows:

[0034]

[0035] The excavation response field correction monitoring data through the surrounding rock includes:

[0036] The absolute displacement Ci is calculated by numerical simulation of each observation point obtained from the excavation response field of the surrounding rock; the observation point Cimin, where the absolute value of the numerically simulated absolute displacement Ci is the smallest, is taken as the displacement reference point imin.

[0037] Based on the multi-point displacement gauge monitoring data obtained on site, the cumulative relative deformation Aimin from the orifice to the displacement reference point imin is obtained, and the absolute deformation B' = Cimin + Aimin is calculated relative to the displacement reference point imin.

[0038] The absolute displacement of the observation point after correction is calculated from the original monitoring point relative to the orifice displacement as A' = B' - Ai; Ai is the relative displacement reading of each observation point relative to the orifice displacement.

[0039] The beneficial effects of this invention are as follows: 1. By combining measured data of relative deformation from different sections of multiple multi-point displacement gauges, optimized surrounding rock parameters based on actual surrounding rock excavation response characteristics are obtained. 2. Reliable surrounding rock excavation response characteristics are obtained through optimized surrounding rock parameters, and on this basis, the field monitoring data are corrected and analyzed. 3. The surrounding rock excavation response characteristics obtained by the above method make full use of the field excavation surrounding rock response monitoring data, which can be used to correct multi-point displacement gauge monitoring data, making the monitoring data obtained by the multi-point displacement gauges more accurate. Attached Figure Description

[0040] Figure 1 The established three-dimensional geological model diagram;

[0041] Figure 2 Deformation analysis diagram monitored by multi-point displacement gauges;

[0042] Figure 3 Deformation distribution map of surrounding rock during tunnel excavation;

[0043] Figure 4 Vector diagram of tunnel excavation displacement. Detailed Implementation

[0044] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the described embodiments of this disclosure without creative effort are within the scope of protection of this disclosure.

[0045] Example 1

[0046] An optimized processing method for monitoring surrounding rock displacement during the excavation of a cavern complex includes the following steps:

[0047] Step 1: Based on the on-site geological survey data, determine the strength parameters of the rock mass and the coordinate data and elevation information of the target area to establish a three-dimensional geological model.

[0048] Modeling: The DEM elevation information of the target area downloaded from satellite-acquired geographic information data is imported into 3D modeling software. In the software, the DEM data is analyzed to extract the surface contour lines (contour lines). Using the extracted contour lines and elevation information, combined with terrain coordinate data, the software generates a 3D mesh. This mesh is then stretched vertically to the corresponding elevation, forming a 3D terrain entity model. The terrain entity model is then segmented and meshed to ultimately generate a 3D geological model suitable for numerical simulation analysis. The entire process progressively transforms 2D graphics into 3D entities, and through detailed mesh generation and manifold mesh combination, ensures the accuracy and usability of the model in numerical simulation. The established 3D geological model is as follows: Figure 1 .

[0049] Determining the strength parameters of the rock mass: The exposed bedrock in this area is mostly granite. For the dikes and alteration zones exposed by the strata, given their complex and discrete distribution, the strength parameters of the rock mass are only considered from the perspective of the surrounding rock category in the three-dimensional geological model of numerical calculation. The strength parameters of the rock mass are obtained through laboratory tests, as shown in Table 1 below: In Table 1, II, III1, III2, IV, and V represent Class II, Class III1, Class III2, Class IV, and Class V surrounding rock, respectively, which are rock masses classified into different quality grades according to the specifications.

[0050] Table 1. Parameters of the surrounding rock of granite

[0051]

[0052] Step 2: Based on the geostress data of the target area obtained from field measurements, the corresponding geostress field is calculated using the multiple linear regression method.

[0053] Considering the tectonic activity of the target region, the following tectonic modes for the formation of the geostress field are considered in the three-dimensional numerical calculation: (a) self-weight, (b) X-direction compressional structure, (c) Y-direction compressional structure, (d) XY horizontal shear structure, (e) vertical shear tectonic factors on the ZY plane and vertical shear structure on the ZX plane.

[0054] The stress generated by the construction modes (a) to (e) is calculated using finite difference numerical simulation software, and the three-dimensional initial geostress of the observation points in the target area is obtained by solving equations (1) to (2).

[0055] Based on the principle of linear superposition under elastic working conditions, the expression for the calculated three-dimensional initial geostress value at the observation point is:

[0056]

[0057] in: The values ​​are calculated based on the regression of j stress components from k observation points, where j∈[1,6] represents the six stress components generated by tectonic modes (a)~(e); L i (i = 1, 2, 3, ..., n) are regression coefficients, n = 6; σ xx σ yy σ xy σ xz σ yz The stresses generated by the construction mode at the k-th observation point in five directions; σ w This represents the stress generated by gravity at the kth observation point; It is the numerical calculation value of the j-th stress component at the k-th observation point under the i-th partial load factor;

[0058] Measured values ​​of ground stress and calculated value The error S between 残 for:

[0059]

[0060] in, It is the measured value of the j-th stress component at the k-th observation point, where j∈[1,6]; the number of stress observation points is m.

[0061] By obtaining the observed values and calculated value Between error S 残 The regression coefficient L obtained when it is at its minimum value i (i = 1, 2, 3, ..., n), n = 6;

[0062] According to the least squares principle, S 残 The equation for the minimum value is:

[0063]

[0064] By solving this system of matrix equations, all regression coefficients can be obtained.

[0065] Step 3: Determine the approximate range of the surrounding rock strength deformation parameters, as shown in Table 2:

[0066] The approximate range of the surrounding rock strength deformation parameters was determined from the rock mechanics parameter manual and laboratory tests, taking into account seismic conditions that take into account lithology. The cohesion and internal friction angle were obtained from laboratory tests.

[0067] Table 2 Range of surrounding rock strength and deformation parameters

[0068] parameter Young's modulus E / GPa Poisson's ratio μ Cohesion c / MPa internal friction angle / ° scope 2-15 0.23-0.30 8-16 20-35

[0069] Step 4: Each rock strength deformation parameter was divided into five different factor levels. 25 orthogonal test groups for rock strength deformation parameters were designed through orthogonal experiments, as shown in Table 3.

[0070] Table 3. Rock strength and deformation parameters of orthogonal experimental design

[0071] Test No. Young's modulus E (GPa) Poisson's ratio μ Cohesion c (MPa) Angle of internal friction (°) 1 2 0.23 8 20 2 2 0.2475 10 23.75 3 2 0.265 12 27.75 4 2 0.2825 14 31.25 5 2 0.30 16 35.00 6 5.25 0.23 10 27.75 7 5.25 0.2475 12 31.25 8 5.25 0.265 14 35.00 9 5.25 0.2825 16 20 10 5.25 0.30 8 23.75 11 8.50 0.23 12 35.00 12 8.50 0.2475 14 20 13 8.50 0.265 16 23.75 14 8.50 0.2825 8 27.75 15 8.50 0.30 10 31.25 16 11.75 0.23 14 23.75 17 11.75 0.2475 16 27.75 18 11.75 0.265 8 31.25 19 11.75 0.2825 10 35.00

[0072] Continued from Table 3

[0073]

[0074] Based on the three-dimensional geostress field of the target area, the corresponding surrounding rock parameters are input into the numerical model. The displacement response field of the surrounding rock is calculated using Flac3D. In Flac3D, the surrounding rock parameters are assigned to the target area using its built-in program, and then the rock mass is excavated and balanced to obtain the displacement response field. The numerical model in this step differs from the stress field in step 2; it is a new model formed after inputting the geostress field data of the target area and the strength and deformation parameters of the surrounding rock. The output result is the displacement response field. Based on the displacement response field, the displacements at various locations of multiple field-deployed multi-point displacement gauges are extracted, and the relative deformation between different locations is calculated. For example, Table 4 shows the numerical simulation relative deformation of different sections of multiple multi-point displacement gauges calculated based on a set of surrounding rock strength and deformation parameters in the geostress field of the target area.

[0075] Table 4. Relative deformation of multiple multi-point displacement gauge sections obtained from the calculation of the surrounding rock strength and deformation parameters of the 6 groups.

[0076]

[0077] like Figure 2 As shown, a multi-point displacement meter has four measuring points. The distance from each measuring point to the orifice is a measuring segment, and a multi-point displacement meter has a total of four measuring segments.

[0078] Based on the displacement response field, the displacements at various locations of multiple field-deployed multi-point displacement gauges are extracted. Assuming that the displacement at the orifice calculated by one multi-point displacement gauge is D0, and the displacement at the r-th measuring point of the multi-point displacement gauge is D... r Then, its relative deformation for the 0-r segment of the multi-point displacement gauge is D0-D. r Therefore, the relative deformation of the test section at different locations can be calculated.

[0079] Step 5: Use the surrounding rock strength deformation parameters of different groups as output data yt, where yt is the data in Table 3; use the corresponding calculated relative deformation of different test sections as input data group xt, where xt is the data in Table 4; use the input data and their corresponding output data as training data, and train a BP neural network to obtain the nonlinear relationship f(x) between the input data and the output data:

[0080]

[0081] The BP neural network has a simple structure, consisting of an input layer, an intermediate layer, and an output layer. In this embodiment, the training data is 25 sets.

[0082] In the formula, x represents the input node representation of the neural network; y represents the output node representation of the neural network; f(v, h1, ..., h) p (w) is the established multi-layer neural network structure, where v is the input layer F. x Number of nodes (assuming the input data set has 6 monitoring data points for multi-point displacement, then v = 6), h1 is the number of elements in the intermediate layer F1, h p For the intermediate layer F p The number of units, w is the output layer F y The number of nodes (assuming the output data set has 4 data points for the surrounding rock strength and deformation parameters, then w = 4), h v R is the number of units in the intermediate layer of the v-th layer; v →R w Represents R v To R w The mapping.

[0083] The root mean square error (RMSE) is selected as the error function. Based on the RMSE of the output layer, the weights and biases of the intermediate layers are adjusted in reverse until the error converges. The error function is the square root of the sum of the squared averages of the differences between all true and predicted values. The calculation formula is as follows:

[0084]

[0085] In this embodiment, the number of intermediate layers is 2, and the number of layer units is 46 and 53.

[0086] The BP neural network was trained to obtain the nonlinear mapping between the model inversion parameters (modulus, cohesion, etc.) and the relative deformation of different sections of different multi-point displacement gauges inside the model, which is the trained BP neural network model.

[0087] Since it is impossible to use the measured model as training data, the goal of this embodiment is to obtain a reliable numerical simulation model. Therefore, training data is first obtained using the surrogate numerical simulation model in Table 4, and then the model parameters (surrounding rock strength and deformation parameters) are obtained by inversion from real monitoring data according to step 6. The ultimate goal is to obtain a reliable numerical model.

[0088] Step 6: After obtaining multi-point displacement gauge monitoring data on-site, the trained BP neural network model is used with the actual relative deformation data of different measurement sections obtained from the multi-point displacement gauges as input data to invert and obtain the optimal surrounding rock strength deformation parameters. The optimal surrounding rock strength deformation parameters obtained by inputting the multi-point displacement gauge monitoring data from this step into the BP neural network model are shown in Table 5:

[0089] Table 5 Range of surrounding rock strength and deformation parameters

[0090]

[0091] Step 7: Input the optimized surrounding rock deformation strength parameters obtained from Step 6 and the three-dimensional geostress field of the target area obtained from Step 2 into the numerical model Flac3D to obtain the excavation response field of the surrounding rock. Excavate the rock mass based on the optimized surrounding rock strength deformation parameters obtained in Step 6 to obtain the excavation response characteristics of the surrounding rock. By comparing the measured surrounding rock excavation response characteristics with the numerical simulation surrounding rock excavation response characteristics, as shown in Table 6, the relaxation depth of the surrounding rock upstream of 0+161.9 and the relative deformation of the M4tys3-1 and M4tys5-2 multi-point displacement gauges before and after excavating the fourth layer are compared. The results are shown in Table 6. Among them, the relaxation depth can also be called the failure depth; 0+161.9 refers to the coordinate position of a certain target area. The numerical simulation surrounding rock excavation response characteristics calculated by the optimized surrounding rock strength deformation parameters are in good agreement with the field measured surrounding rock excavation response characteristics.

[0092] Table 6 Comparison of numerical simulation values ​​and field measured values ​​of inverted optimal surrounding rock strength and deformation parameters

[0093] Location Monitoring categories Numerical simulation values On-site measured values Adjusting the horizontal direction 0+161.9 upstream side Rock relaxation depth 1m 1.2m <![CDATA[M 4 tys3-1 Multi-point displacement meter Excavation of the fourth layer relative deformation 0.375mm 0.40mm <![CDATA[M 4 tys5-2 Multi-point displacement meter Excavation of the fourth layer relative deformation 0.25 0.2

[0094] Step 8: Based on the multi-point displacement gauge monitoring data obtained on-site, the monitoring data is corrected by numerical simulation of the surrounding rock excavation response characteristics. The following is referred to as M... 4 ZB3-2-1 ~M 4 ZB3-2-4 Taking a multi-point displacement meter as an example, the specific calibration steps are explained below:

[0095] Analysis of monitoring data revealed a significant displacement change at a location 17m deep where a multi-point displacement gauge (total length 34m) was buried after excavation, as shown in Table 7. Direct analysis of the monitoring data also indicates that... Figure 2 As shown: (1) The first point M at a depth of 2m from the orifice. 4 ZB3-2-1 (1) The deformation is about 11 mm; (2) The deformation is 7-11 = -4 mm in the range of 2-5 m of surrounding rock depth, and 2-7 = -5 mm in the range of 5-17 m of surrounding rock depth; (3) The deformation is 20-2 mm = 18 mm in the range of 17-34 m of surrounding rock depth.

[0096] Table 7. Monitoring data from multi-point displacement gauges

[0097] Measurement point number Cumulative relative deformation <![CDATA[M 4 ZB3-2-1 ]]> 11.10 <![CDATA[M 4 ZB3-2-2 ]]> 7.02 <![CDATA[M 4 ZB3-2-3 ]]> 2.33 <![CDATA[M 4 ZB3-2-4 ]]> 20.53

[0098] The rationality of the monitoring data is analyzed as follows:

[0099] like Figure 3 As shown, based on the deformation cloud map of the current excavation stage of the main transformer room, combined with the location of the displacement observation points, it can be seen that the deepest observation point at 34m is close to the tailgate sidewall. From the borehole opening to the 34m depth, the displacement at each observation point shows a trend of deformation first decreasing and then increasing. Further analysis using the displacement vector diagram, as shown... Figure 4 As shown in the figure, analysis reveals that the displacement vector direction from the observation point orifice to 17m points towards the main transformer chamber, while the displacement vector direction from 17m to 34m points towards the tailgate chamber. Currently, the tailgate chamber is undergoing the third layer pre-splitting process, which differs from the calculation stage, but the deformation vector direction follows the same pattern.

[0100] Therefore, the deformation measured at the fourth observation point was 20 mm, which may be related to the deformation of the last point of the multi-point displacement gauge towards the main transformer chamber. Thus, this deformation is likely substantial; considering the orifice deformation reached 11 mm, the measured deformation of 20 mm is reasonable.

[0101] The following correction is based on the simulation data: The absolute displacement C of each observation point obtained by numerical simulation is calculated. The point with the smallest absolute value of the numerical simulation absolute displacement C is taken as the displacement reference point. The numerical simulation absolute displacement at a depth of 5m is C = 7.5mm. As the displacement reference point, the relative orifice deformation at a depth of 5m is A2 = 7.02mm.

[0102] The absolute displacement of the orifice equals the absolute displacement of the point to be corrected (a) plus the relative displacement of the section from point a to the orifice. Therefore, the absolute deformation at the orifice after correction is B' = C + A2 = 14.52 mm. The absolute displacement of the observation point after correction can be calculated from the original observation point's displacement relative to the orifice as A' = B' - Ai, where Ai is the relative displacement reading of each observation point relative to the orifice. The corrected absolute displacements of the remaining observation points are derived sequentially, as shown in Table 8. It can be seen that the absolute displacement of the observation points after correction is significantly closer to the absolute displacement calculated by numerical simulation. From the above results, it can be seen that there is a certain error when using the absolute displacement of the original observation points at different base points for the multi-point displacement gauge. Therefore, it is considered that using the relative displacements of multiple observation points for numerical simulation parameter inversion is more accurate than using single-point multi-point displacement.

[0103] The absolute deformation at the orifice after correction is B', the absolute displacement of the observation point after correction is A', and the original monitoring relative orifice displacement Ai is the actual monitoring value.

[0104] The absolute displacement of the original observation point is calculated by taking the deepest measuring point as a fixed point according to the original method. For example, if the borehole opening is 20.53, then the absolute displacement at 2m is 20.53 - 11.10.

[0105] The absolute displacement of the observation point after correction is as follows: the absolute displacement at the orifice was first corrected to 14.52 mm. Then the absolute displacement at the observation point = the absolute displacement at the orifice - the original displacement of the observation point relative to the orifice. For example, at 2 m, it is 14.52 - 11.10 = 3.42 mm.

[0106] The absolute displacement calculated by numerical simulation is the absolute displacement directly obtained through numerical simulation in step 7.

[0107] Table 8. Displacements of measuring points obtained from numerical simulation.

[0108]

[0109] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for optimizing the processing of surrounding rock displacement monitoring data during the excavation of a cavern complex, characterized in that, Includes the following steps: Step S1: Based on the on-site geological survey data, determine the strength parameters of the rock mass and the coordinate data and elevation information of the target area to establish a three-dimensional geological model; Step S2: Based on the geostress data of the observation points in the target area obtained by field measurement, the corresponding three-dimensional geostress field is calculated by multiple linear regression method; Step S3: Obtain the range of surrounding rock strength deformation parameters. The surrounding rock strength deformation parameters include Young's modulus, Poisson's ratio, cohesion, and internal friction angle. Each surrounding rock strength deformation parameter is divided into t different factor levels with the same number. t groups of orthogonal test groups of surrounding rock strength deformation parameters are designed through orthogonal test. Step S4: Input the three-dimensional geostress field of the target area and the surrounding rock strength deformation parameters of different groups into the numerical model, calculate the displacement response field of the surrounding rock, extract the displacement at each location of multiple field-deployed multi-point displacement gauges based on the displacement response field, and calculate the relative deformation of the measuring segments between different locations. Step S5: Train the BP neural network using the nonlinear relationship between the surrounding rock strength deformation parameters calculated in step S4 and the corresponding relative deformation of different test sections to obtain the BP neural network model. Input the actual relative deformation of different test sections obtained by multi-point displacement gauge monitoring into the BP neural network model to invert and obtain the optimal surrounding rock strength deformation parameters. Step S6: Input the preferred surrounding rock deformation strength parameters and the three-dimensional geostress field obtained in step S2 into the numerical model to calculate the excavation response field of the surrounding rock. Step S7: Based on the multi-point displacement gauge monitoring data obtained on site, the absolute displacement Ci is calculated by numerical simulation of each observation point obtained through the excavation response field of the surrounding rock; the observation point Cimin, where the absolute value of the numerically simulated absolute displacement Ci is the smallest, is taken as the displacement reference point imin. Based on the multi-point displacement gauge monitoring data obtained on site, the cumulative relative deformation Aimin from the orifice to the displacement reference point imin is obtained, and the absolute deformation B' = Cimin + Aimin is calculated relative to the displacement reference point imin. The absolute displacement of the observation point after correction is calculated from the original monitoring point's relative displacement to the orifice: A' = B' - Ai; Ai is the relative displacement reading of each observation point relative to the orifice. In step S4, the displacement of the orifice calculated by a multi-point displacement gauge is D0, and the displacement of the r-th observation point of the multi-point displacement gauge is D. r Then, its relative deformation for the 0th to rth measurement segments of the multi-point displacement gauge is D0-D. r ; Step S5, training the BP neural network, includes: The relative deformation of different measuring sections of the multi-point displacement gauge calculated in step S4 is used as the input data set xt, and the surrounding rock deformation strength parameters of the different sets corresponding to it in step S4 are used as the output data yt. The input data and output data are used as training data, and the nonlinear relationship f(x) between the input data and output data is obtained by training a BP neural network: （4）； In the formula, x represents the input node representation of the neural network; y represents the output node representation of the neural network. It is a multi-layer neural network structure, where v is the input layer F. x The number of nodes, h1 is the number of units in the intermediate layer F1, h p For the intermediate layer F p The number of units, w is the output layer F y The number of nodes, h v Let v be the number of units in the intermediate layer of the v-th layer; Represents R v To R w Mapping; The root mean square error (RMSE) is selected as the error function. Based on the RMSE of the output layer, the weights and biases of the intermediate layers are adjusted in reverse until the error converges. The error function is the square root of the sum of the squared averages of the differences between all true and predicted values. The calculation formula is as follows: (5); Among them, y i and These are the true and predicted values ​​of the surrounding rock deformation strength parameters in different sections of the multi-point displacement gauge.

2. The method according to claim 1, characterized in that, Step S1, the modeling process, includes: Step 11: Download the DEM elevation information of the target area from the satellite-acquired geographic information data and import it into the 3D modeling software; Step 12: Analyze the DEM data to extract the outline of the land surface. Using the extracted outline and elevation information, combined with the terrain coordinate data, the software generates a three-dimensional mesh. Step 13: Stretch the generated 3D mesh along the vertical direction to the corresponding elevation to form a 3D terrain entity model. Then, cut and mesh the terrain entity model to generate a 3D geological model suitable for numerical simulation analysis.

3. The method according to claim 1, characterized in that, Step S2 includes: Step 21: Construct the geostress field. The tectonic modes include: (a) self-weight, (b) X-direction compression, (c) Y-direction compression, (d) XY horizontal shear, (e) vertical shear tectonic factors on the ZY plane and vertical shear tectonic factors on the ZX plane. Calculate the stress generated by the tectonic modes (a) to (e) using finite difference numerical simulation software. Step 22: Using the principle of linear superposition under elastic working conditions, calculate the initial ground stress value: (1) in: The regression calculation values ​​of j stress components based on k observation points are given, where j∈[1,6] represents the 6 stress components generated by the construction modes (a)~(e); These are the regression coefficients, n=6; The stress generated at the k-th observation point in five directions by the construction mode; This represents the stress generated by gravity at the kth observation point; It is the numerical calculation value of the j-th stress component at the k-th observation point under the i-th partial load factor; Step 23: Calculate the measured values ​​of geostress and calculated value Error between: (2) in, It is the measured value of the j-th stress component at the k-th observation point, where j∈[1,6]; the number of stress observation points is m; Step 24: Obtain the observed values and calculated value The regression coefficient obtained by minimizing the error between them n=6; According to the least squares principle, this makes The system of equations that yields the minimum value is: (3)； By solving this system of matrix equations, all regression coefficients can be obtained.

Images