A method for predicting lining displacement curves in overlapping tunnel excavation

Abstract

The present invention discloses a method for predicting displacement curves of linings in cross-overlap tunnel excavation, the method comprising: obtaining multiple tunnel variables between an existing tunnel and a newly built tunnel under target working conditions, inputting the multiple tunnel variables into a trained generalized regression neural network, and outputting multiple specific parameters, wherein training data is generated based on a two-dimensional finite element model, and a generalized regression neural network model is trained according to the training data; all the specific parameters are substituted into a pre-constructed displacement equation to obtain a displacement equation of the existing tunnel; and a lining displacement curve is generated according to the displacement equation. The present invention can quickly generate an accurate displacement curve.

Description

Technical Field

[0001] The invention relates to the technical field of tunnel engineering, and in particular to a method for predicting displacement curves of linings in cross-overlap tunnel excavation. Background Art

[0002] Shield tunnels have been widely used in subways, urban expressways, highways and other projects. With the development of urbanization and the improvement of infrastructure construction, it is becoming more and more common for two or more parallel tunnels to exist at close range. In two or more parallel tunnels, for the safety of the tunnels, it is necessary to obtain the displacement information of the tunnel lining.

[0003] However, the current data-driven approach based on machine learning is based on the prediction of single-line tunnels when predicting the lining displacement caused by overlapping tunnels. It only focuses on the maximum displacement or the displacement at a specific location. Therefore, it is impossible to accurately predict the displacement curve of the cross-overlapping tunnel excavation lining of overlapping tunnels.

[0004] Therefore, the prior art still needs to be improved and developed. Summary of the invention

[0005] The main purpose of the present invention is to provide a method, system, terminal and computer-readable storage medium for predicting the displacement curve of the lining of overlapping tunnel excavation, aiming to solve the problem that the data-driven method with machine learning as the core in the prior art predicts the lining displacement caused by overlapping tunnels, which is the prediction of single-line tunnels and only focuses on the maximum displacement or the displacement at a specific position. Therefore, the prediction of the displacement curve of the lining of overlapping tunnel excavation cannot achieve accurate prediction effect.

[0006] To achieve the above object, the present invention provides a method for predicting displacement curves of linings in cross-overlapping tunnel excavation, the method comprising the following steps:

[0007] Acquire multiple tunnel variables between the existing tunnel and the newly built tunnel under target working conditions, input the multiple tunnel variables into a trained generalized regression neural network, and output multiple specific parameters, wherein training data is generated based on a two-dimensional finite element model, and the generalized regression neural network model is trained according to the training data;

[0008] Substituting all the specific parameters into a pre-constructed displacement equation to obtain the displacement equation of the existing tunnel;

[0009] According to the displacement equation, a lining displacement curve is generated.

[0010] Optionally, the training process of the generalized regression neural network specifically includes:

[0011] Generating the training data based on the two-dimensional finite element model, and dividing the training data into a training set and a validation set;

[0012] The generalized regression neural network is trained according to the training set, wherein the kernel width is optimized according to the grid search method and the cross-validation method in each training, and the corresponding determination coefficient is calculated after each training, and when the determination coefficient reaches the preset requirement, the training is terminated to obtain a preliminarily trained generalized regression neural network;

[0013] Inputting the verification set into a preliminarily trained generalized regression neural network, outputting a verification result, generating a first verification displacement curve according to the verification result, and generating a comparison result according to the verification set and the first verification displacement curve;

[0014] When the comparison result meets the preset comparison requirement, the current generalized regression neural network is used as the trained generalized regression neural network.

[0015] Optionally, generating the training data based on the two-dimensional finite element model and dividing the training data into a training set and a validation set specifically includes:

[0016] Acquire training tunnel variables, wherein the training tunnel variables include training tunnel diameter, training tunnel burial depth, training stratum loss rate and tunnel relative position;

[0017] constructing a two-dimensional tunnel model in the two-dimensional finite element model based on a training tunnel project and the training tunnel variables;

[0018] The tunnel construction process is simulated in the two-dimensional finite element model to obtain the training specific parameters corresponding to the training tunnel variables, the training data is constructed according to the training specific parameters corresponding to the training tunnel variables, and the training data is divided into a training set and a validation set.

[0019] Optionally, the step of inputting the verification set into a preliminarily trained generalized regression neural network, outputting a verification result, generating a first verification displacement curve according to the verification result, and generating a comparison result according to the verification set and the first verification displacement curve specifically includes:

[0020] Inputting the verification set into a preliminarily trained generalized regression neural network and outputting a verification result;

[0021] Substituting the verification result into a pre-constructed displacement equation to generate a plurality of training displacements, generating a plurality of first verification displacement curves according to the plurality of training displacement equations, and generating a plurality of second verification displacement curves according to the training specific parameters corresponding to each training tunnel variable in the verification set;

[0022] A comparison result is generated according to all of the first verification displacement curves and the corresponding second verification displacement curves.

[0023] Optionally, the step of obtaining a plurality of tunnel variables between the existing tunnel and the newly built tunnel under target working conditions, inputting the plurality of tunnel variables into a trained generalized regression neural network, and outputting a plurality of specific parameters specifically includes:

[0024] Acquire multiple tunnel variables between the existing tunnel and the newly built tunnel under target working conditions, wherein the tunnel variables include tunnel diameter, tunnel burial depth, stratum loss rate and tunnel relative position, and the tunnel relative position in the tunnel variables under different working conditions is different;

[0025] The plurality of tunnel variables under the target working condition are input into the trained generalized regression neural network, and a plurality of specific parameters are output, wherein each group of specific parameters includes a first elliptical convergence, a second elliptical convergence, an overall horizontal displacement, an overall vertical displacement and an overall displacement deflection.

[0026] Optionally, the step of substituting all the specific parameters into a pre-constructed displacement equation to obtain the lining displacement equation of the existing tunnel further includes:

[0027] Presetting the parameter representation of the first elliptical convergence, the second elliptical convergence, the overall horizontal displacement, the overall vertical displacement and the overall displacement deflection;

[0028] The lining displacement equations are generated based on the parametric representation of the first elliptical convergence, the second elliptical convergence, the overall horizontal displacement, the overall vertical displacement and the overall displacement deflection.

[0029] Optionally, substituting all the specific parameters into a pre-constructed displacement equation to obtain the displacement equation of the existing tunnel specifically includes:

[0030] Obtain the target demand and substitute all the specific parameters into the pre-built lining displacement equation;

[0031] The lining displacement equation after substituting the parameters is calculated to obtain the target equation, and the target equation is converted into the displacement equation according to the target requirements.

[0032] In addition, to achieve the above-mentioned purpose, the present invention also provides a cross-overlap tunnel excavation lining displacement curve prediction system, wherein the cross-overlap tunnel excavation lining displacement curve prediction system comprises:

[0033] A specific parameter output module is used to obtain a plurality of tunnel variables between the existing tunnel and the newly built tunnel under target working conditions, input the plurality of tunnel variables into a trained generalized regression neural network, and output a plurality of specific parameters, wherein training data is generated based on a two-dimensional finite element model, and a generalized regression neural network model is trained based on the training data;

[0034] A displacement generating module, used for substituting all the specific parameters into a pre-constructed displacement equation to obtain a displacement equation of the existing tunnel;

[0035] The result output module is used to generate a lining displacement curve according to the displacement equation.

[0036] In addition, to achieve the above-mentioned purpose, the present invention also provides a terminal, wherein the terminal includes: a memory, a processor, and a cross-overlapping tunnel excavation lining displacement curve prediction program stored in the memory and runnable on the processor, wherein the cross-overlapping tunnel excavation lining displacement curve prediction program, when executed by the processor, implements the steps of the cross-overlapping tunnel excavation lining displacement curve prediction method as described above.

[0037] In addition, to achieve the above-mentioned purpose, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a cross-overlapping tunnel excavation lining displacement curve prediction program, and when the cross-overlapping tunnel excavation lining displacement curve prediction program is executed by a processor, the steps of the cross-overlapping tunnel excavation lining displacement curve prediction method as described above are implemented.

[0038] In the present invention, multiple tunnel variables between the existing tunnel and the newly built tunnel under the target working conditions are obtained, and the multiple tunnel variables are input into the trained generalized regression neural network to output multiple specific parameters, wherein training data is generated based on a two-dimensional finite element model, and the generalized regression neural network model is trained according to the training data; all the specific parameters are substituted into a pre-constructed displacement equation to obtain the displacement equation of the existing tunnel; and a lining displacement curve is generated according to the displacement equation. The present invention combines theoretical analysis with a neural network algorithm to decompose the lining displacement of the existing tunnel into four items and quantify it using five specific parameters, thereby deriving the lining displacement, that is, substituting the parameters generated by the neural network into the corresponding displacement equation to obtain the corresponding displacement, thereby generating an accurate displacement curve. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a flow chart of a preferred embodiment of the method for predicting displacement curves of linings in cross-overlap tunnel excavation of the present invention;

[0040] Figure 2It is a characteristic schematic diagram of the displacement curve of an existing tunnel in the method for predicting the displacement curve of the lining of the cross-overlapping tunnel excavation of the present invention;

[0041] Figure 3 It is a schematic diagram of the decomposition of the lining displacement of an existing tunnel in the method for predicting the lining displacement curve of the cross-overlapping tunnel excavation of the present invention;

[0042] Figure 4 It is a schematic diagram of the generalized regression neural network structure in the method for predicting the displacement curve of the lining in the cross-overlapping tunnel excavation of the present invention;

[0043] Figure 5 It is a schematic diagram of the meaning of tunnel variables in the method for predicting lining displacement curves of cross-overlap tunnel excavation of the present invention;

[0044] Figure 6 It is a schematic diagram of the technical route of the method for predicting the displacement curve of the lining in the cross-overlap tunnel excavation of the present invention;

[0045] Figure 7 It is a structural diagram of a preferred embodiment of the cross-overlap tunnel excavation lining displacement curve prediction system of the present invention;

[0046] Figure 8 It is a structural diagram of a preferred embodiment of the terminal of the present invention. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solution and advantages of the present invention clearer and more specific, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0048] Shield tunnels have been widely used in subways, urban expressways, highways and other projects. With the development of urbanization and the improvement of infrastructure construction, it is becoming more and more common for two or more parallel tunnels to exist at close distances. Two or more parallel tunnels are called "parallel overlapping tunnels", or "overlapping tunnels" for short. Tunnels with earlier construction periods are called existing tunnels (ET for short), and tunnels with later construction periods are called new tunnels (NT for short). In overlapping tunnels, the lining displacement of existing tunnels exhibits different properties from single-line tunnels, and the mechanism of mutual influence between tunnels is still unclear, making prediction more difficult. In two or more parallel tunnels, for the safety of the tunnels, it is necessary to obtain the corresponding tunnel lining displacement information.

[0049] At present, there are two main ways to predict the lining displacement caused by overlapping tunnels: one is the knowledge-driven method with theoretical analysis as the core; the other is the data-driven method with machine learning as the core. Regarding knowledge-driven, the lining displacement will be decomposed to simplify the mathematical model, but the form of the analytical solution is still relatively complex. Although these analytical solutions conform to the principles of mathematics and mechanics and have good interpretability, they are often based on idealized assumptions and are too complex in form, which is not conducive to direct application in actual engineering. Regarding data-driven, since the construction of new tunnels in overlapping tunnels will have an impact on existing tunnels, the lining displacement of existing tunnels is often not elliptical like single-line tunnels, but irregular in shape. Therefore, the prediction of the lining displacement curve of the cross-overlapping tunnel excavation of overlapping tunnels cannot achieve accurate prediction results.

[0050] In response to one or more of the above problems, the present invention obtains multiple tunnel variables between the existing tunnel and the newly built tunnel under target working conditions, inputs the multiple tunnel variables into a trained generalized regression neural network, and outputs multiple specific parameters, wherein training data is generated based on a two-dimensional finite element model, and a generalized regression neural network model is trained according to the training data; all of the specific parameters are substituted into a pre-constructed displacement equation to obtain the displacement equation of the existing tunnel; and a lining displacement curve is generated according to the displacement equation.

[0051] The method for predicting displacement curves of linings in cross-overlap tunnel excavation according to a preferred embodiment of the present invention is as follows: Figure 1 As shown, the cross-overlap tunnel excavation lining displacement curve prediction method comprises the following steps:

[0052] Step S10, obtaining multiple tunnel variables between the existing tunnel and the newly built tunnel under the target working conditions, inputting the multiple tunnel variables into the trained generalized regression neural network, and outputting multiple specific parameters, wherein training data is generated based on the two-dimensional finite element model, and the generalized regression neural network model is trained according to the training data.

[0053] It should be noted that the existing tunnel is an existing tunnel in the two parallel tunnels, and the new tunnel is a tunnel to be constructed. In the present invention, the corresponding variables between the existing tunnel and the new tunnel under different working conditions are first obtained, and then the corresponding variables are input into the trained generalized regression neural network, and the corresponding multiple specific parameters are output, wherein the multiple specific parameters include 5 parameters for quantifying displacement characteristics.

[0054] Furthermore, the method of substituting all the specific parameters into the pre-constructed displacement equation to obtain the lining displacement equation of the existing tunnel also includes:

[0055] Presetting the parameter representation of the first elliptical convergence, the second elliptical convergence, the overall horizontal displacement, the overall vertical displacement and the overall displacement deflection;

[0056] The lining displacement equations are generated based on the parametric representation of the first elliptical convergence, the second elliptical convergence, the overall horizontal displacement, the overall vertical displacement and the overall displacement deflection.

[0057] The present invention decomposes the lining displacement of the existing tunnel into four parts: elliptical convergence, overall horizontal displacement, overall vertical displacement and displacement deflection. These four parts correspond to different deformation characteristics of the tunnel lining and are expressed by five parameters (the first elliptical convergence , the second ellipticization converges , overall horizontal displacement , overall vertical displacement , overall displacement deflection ) to quantify the displacement characteristics of each part.

[0058] Specifically, in the present invention, since the lining displacement of the cross-overlapping tunnel is different from that of the single-line shield tunnel, the displacement deformation of the existing tunnel is different from that of the single-line tunnel in the following two aspects: the excavation of the new tunnel causes the existing tunnel to have a horizontal displacement; due to the excavation of the new tunnel, the lining displacement is deflected, that is, the maximum convergence of the tunnel is no longer on the horizontal axis or the vertical axis. Correspondingly, it can also be obtained through simulation, such as Figure 2 As shown, in the cross-overlapping tunnels, the black dotted line represents the situation when the ET lining deformation is magnified 200 times during the construction of the new tunnel NT.

[0059] The displacement of each point on the lining of a single-line shield tunnel can be regarded as the superposition of two displacements: the first is the radial deformation of the tunnel toward the center point. , that is, the ellipticization converges, yes The function of is the corresponding polar angle in polar coordinates; the second is the overall vertical displacement of the tunnel , which is the same at all points in the tunnel. Therefore, based on the decomposition and calculation formula of the lining displacement of the single-line shield tunnel according to the overlapping tunnel, the following two assumptions are added: Due to the construction of the new tunnel, the existing tunnel has an overall horizontal displacement ; Due to the construction of the new tunnel, the existing tunnel converges to an ellipse at a certain polar angle This produces an angular rotation, becoming Based on the above analysis, in the prediction of the displacement of the existing tunnel lining, the displacement is Figure 3 The decomposition is shown as follows: Elliptic convergence ( ), describing the characteristics of tunnel convergence deformation, which is distributed in an elliptical shape; the overall horizontal displacement ( ), which represents the horizontal displacement of the existing tunnel caused by the excavation of the new tunnel; the overall vertical displacement ( ), which indicates the vertical displacement of the entire tunnel; the overall displacement deflection ( ), describing the deflection in displacement direction due to the influence of the new tunnel.

[0060] Correspondingly, the lining displacement of ET The following formula (1) represents:

[0061] ; (1)

[0062] in, is the vector representation of the lining displacement, It is the vector representation corresponding to the angle position generated by the radial deformation of the tunnel at a certain polar angle. is the vector representation of the overall vertical displacement, is the vector representation of the overall horizontal displacement. Therefore, for an existing tunnel, its radial displacement With tangential displacement It can be expressed as the following formula (2):

[0063] ; (2)

[0064] The horizontal displacement With vertical displacement It is expressed as the following formula (3):

[0065] ; (3)

[0066] The above formulas (1), (2) and (3) can all represent the pre-constructed displacement equations of existing tunnels, where formula (1) is the total formula of lining displacement, and formula (2) is the formula decomposed from formula (1) in the polar coordinate system. With tangential displacement The corresponding lining displacement can be obtained by superposition , Formula (3) is the formula (1) decomposed in the rectangular coordinate system, which is obtained by the horizontal displacement With vertical displacement The corresponding lining displacement can be obtained by superposition.

[0067] Therefore, it is only necessary to determine the convergence of the first ellipticization , the second ellipticization converges , overall horizontal displacement , overall vertical displacement and overall displacement deflection The values ​​of these five specific parameters can be used to obtain the lining displacement of each point on the existing tunnel.

[0068] Among them, the elliptical convergence is mainly related to the excavation of the existing tunnel itself, and has little influence on the newly built tunnel; the overall horizontal displacement is closely related to the horizontal distance Δx between the tunnels; the overall vertical displacement is significantly affected by the vertical distance Δy between the newly built tunnel and the existing tunnel; the displacement deflection is mainly affected by the relative position of the two tunnels and has no significant correlation with other variables.

[0069] In the present invention, the parameters of the first elliptical convergence, the second elliptical convergence, the overall horizontal displacement, the overall vertical displacement and the overall displacement deflection are preset, and the corresponding equations in formulas (1), (2) and (3) are obtained.

[0070] Furthermore, the training process of the generalized regression neural network specifically includes:

[0071] Generating the training data based on the two-dimensional finite element model, and dividing the training data into a training set and a validation set;

[0072] The generalized regression neural network is trained according to the training set, wherein the kernel width is optimized according to the grid search method and the cross-validation method in each training, and the corresponding determination coefficient is calculated after each training, and when the determination coefficient reaches the preset requirement, the training is terminated to obtain a preliminarily trained generalized regression neural network;

[0073] Inputting the verification set into a preliminarily trained generalized regression neural network, outputting a verification result, generating a first verification displacement curve according to the verification result, and generating a comparison result according to the verification set and the first verification displacement curve;

[0074] When the comparison result meets the preset comparison requirement, the current generalized regression neural network is used as the trained generalized regression neural network.

[0075] Specifically, the training data is generated by a two-dimensional finite element model, and the training data is divided into a training set and a validation set, so as to be used for training the corresponding generalized regression neural network, and the validation set is used for verification after the training is completed.

[0076] Furthermore, generating the training data based on the two-dimensional finite element model and dividing the training data into a training set and a validation set specifically includes:

[0077] Acquire training tunnel variables, wherein the training tunnel variables include training tunnel diameter, training tunnel burial depth, training stratum loss rate and tunnel relative position;

[0078] constructing a two-dimensional tunnel model in the two-dimensional finite element model based on a training tunnel project and the training tunnel variables;

[0079] The tunnel construction process is simulated in the two-dimensional finite element model to obtain the training specific parameters corresponding to the training tunnel variables, the training data is constructed according to the training specific parameters corresponding to the training tunnel variables, and the training data is divided into a training set and a validation set.

[0080] Specifically, when acquiring training data, the present invention acquires corresponding tunnel variables in overlapping tunnel projects in historical data, wherein the variable data includes training tunnel diameter, training tunnel burial depth, training stratum loss rate and tunnel relative position; in one embodiment of the present invention, the training tunnel diameter can have corresponding values ​​of 6m, 9m, 12m and 15m; the training tunnel burial depth can be set to 1 to 4 times of the tunnel diameter; in the training stratum loss rate, the stratum loss rate combination of the existing tunnel and the newly built tunnel can be set to 1% and 2% respectively, and the combination can be divided into (1%, 1%), (1%, 2%), (2%, 1%) and (2%, 2%); and the relative position of the training tunnel, i.e. the horizontal and vertical distances between the simulated tunnels, has a value range within 5 times of the tunnel diameter, and a set of working conditions is set for every 0.5 times. Therefore, multiple corresponding sets of training tunnel variables can be obtained.

[0081] After obtaining multiple sets of training tunnel variables, combined with the specific training tunnel project, a two-dimensional tunnel model is constructed in the two-dimensional finite element model, and then the tunnel construction process is simulated, wherein the simulated tunnel construction process includes the following steps: Step 1 is to activate the plate unit, freeze the soil, and simulate the excavation of the existing tunnel tunneling machine; Step 2 is to advance the shield machine to the shield tail, and the existing tunnel is subjected to linear contraction; Step 3 is to install the lining, and the plate unit is replaced with the lining; Steps 4 to 6 are to repeat the excavation, advancement and lining installation process of the new tunnel.

[0082] By simulating the tunnel construction process, the lining displacement curve under each working condition can be obtained. By fitting the lining displacement curve, the training specific parameters corresponding to the training tunnel variables can be obtained, that is, the five undetermined parameters corresponding to the training data, namely, the first elliptical convergence, the second elliptical convergence, the overall horizontal displacement, the overall vertical displacement and the overall displacement deflection corresponding to the training data.

[0083] Furthermore, in the process of simulating tunnel construction, after obtaining the corresponding lining displacement curve under each working condition, when fitting, the fitting process adjusts the five parameters in the pre-built displacement equation through the least squares method to minimize the error between the displacement curve predicted by the model and the actual finite element data. Through this fitting method, the parameters of the displacement curve of each tunnel lining can be obtained, thereby completing the establishment of the data set. The determination coefficient of the error between the displacement curve predicted by the model and the actual finite element data R 2 When the pre-set requirements are met, it is considered that the training-specific parameters at this time can be used to generate training data.

[0084] in, R 2 The definition of is formula (4):

[0085] ; (4)

[0086] in, Indicates the first displacement curve of the lining under the current working condition. i The predicted value of the radial displacement of the polar angle is Indicates the first displacement curve of the lining under the current working condition. i The true value of the radial displacement of the polar angle, It represents the average value of all polar radial displacements on the lining displacement curve under the current working condition. represents the total number of polar angles, Can be set by individual users.

[0087] In the fitting process, according to R 2 Based on the minimum principle, the five parameters of each lining curve in the finite element model corresponding to the training tunnel variables are output, thus completing the construction of the data set.

[0088] The training specific parameters corresponding to the training tunnel variables are used as the labels of the training tunnel variables, the corresponding training data are constructed, and the data are divided using the user-set ratio to obtain the corresponding training set and data set.

[0089] Then, the generalized regression neural network is trained according to the training set. The generalized regression neural network (GRNN) in the present invention is a neural network based on radial basis function, which has strong nonlinear mapping ability and is particularly suitable for complex nonlinear prediction tasks under multiple working conditions. The GRNN includes four layers: input layer, pattern layer, summation layer and output layer. Figure 4 The number of nodes in the input layer is the same as the number of input variables, which plays the role of transferring variables. The input is , where X represents the input of the input layer, that is, the tunnel variable corresponding to the working condition. In particular, corresponding to the present invention, the input corresponding to the input layer is the tunnel diameter , burial depth , the ground loss rate of existing tunnels , the ground loss rate of the new tunnel , horizontal distance and vertical distance , etc., as the input of the network; the number of nodes in the pattern layer is consistent with the number of samples in the data set, and each node in the pattern layer calculates the Euclidean distance between the input vector and the sample, and then generates a response value.

[0090] The summation layer contains two nodes for weighted summation of the responses of the pattern layer nodes, one node calculates the weighted sum of all pattern layer outputs, and the other node calculates the denominator of the weighted sum.

[0091] The output layer adds the output of the summation layer and the output is , correspondingly, the output layer of the present invention outputs 5 parameters respectively ( , , , , ), that is, arrive , these parameters are used to describe the lining displacement curve of an existing tunnel.

[0092] In the present invention, the training set is input into the GRNN model for training, so that the model forms a nonlinear mapping by learning the relationship between the input variables and the output parameters; the key parameter of GRNN is the width of the kernel function σ , which affects the response of the pattern layer nodes, is optimized using grid search and cross-validation methods. σ value to ensure the generalization ability and prediction accuracy of the model; and after each training, calculate the corresponding determination coefficient ( R² ), when the determination coefficient reaches the preset requirement, σ When the best prediction effect is achieved, the training is ended and a preliminary trained generalized regression neural network is obtained, in which the threshold of the coefficient of determination set by the user is preset.

[0093] Further, the step of inputting the verification set into a preliminarily trained generalized regression neural network, outputting a verification result, generating a first verification displacement curve according to the verification result, and generating a comparison result according to the verification set and the first verification displacement curve specifically includes:

[0094] Inputting the verification set into a preliminarily trained generalized regression neural network and outputting a verification result;

[0095] Substituting the verification result into a pre-constructed displacement equation to generate a plurality of training displacements, generating a plurality of first verification displacement curves according to the plurality of training displacement equations, and generating a plurality of second verification displacement curves according to the training specific parameters corresponding to each training tunnel variable in the verification set;

[0096] A comparison result is generated according to all of the first verification displacement curves and the corresponding second verification displacement curves.

[0097] Specifically, in the present invention, after obtaining a preliminary trained generalized regression neural network, the corresponding training tunnel variables in the verification set are input into the preliminary trained generalized regression neural network, and the verification result is output. The corresponding verification result is the corresponding training specific parameter. These parameters are substituted into the pre-constructed displacement equation to generate multiple training displacements, and the corresponding first verification displacement curve is generated according to the multiple training displacements; in addition, the second verification displacement curve is generated by the training specific parameters corresponding to the training tunnel variables in the verification set. A comparison result is generated according to all the first verification displacement curves and the corresponding second verification displacement curves, and it is judged whether the comparison result meets the preset comparison requirements. If it meets the requirements, the current generalized regression neural network is used as the trained generalized regression neural network; if it does not meet the requirements, it is trained again. Among them, the preset comparison requirements are the threshold values ​​of the corresponding determination coefficients, and the comparison process is to calculate the determination coefficients and to judge whether the determination coefficients meet the preset comparison requirements.

[0098] Furthermore, the method of obtaining a plurality of tunnel variables between the existing tunnel and the newly built tunnel under the target working condition, inputting the plurality of tunnel variables into the trained generalized regression neural network, and outputting a plurality of specific parameters specifically includes:

[0099] Acquire multiple tunnel variables between the existing tunnel and the newly built tunnel under target working conditions, wherein the tunnel variables include tunnel diameter, tunnel burial depth, stratum loss rate and tunnel relative position, and the tunnel relative position in the tunnel variables under different working conditions is different;

[0100] The plurality of tunnel variables under the target working condition are input into the trained generalized regression neural network, and a plurality of specific parameters are output, wherein each group of specific parameters includes a first elliptical convergence, a second elliptical convergence, an overall horizontal displacement, an overall vertical displacement and an overall displacement deflection.

[0101] Specifically, multiple tunnel variables between the existing tunnel and the newly built tunnel under different working conditions are obtained, wherein each tunnel variable includes tunnel diameter D, tunnel burial depth C, stratum loss rate and tunnel relative position, the tunnel relative position includes the horizontal and vertical distances Δx and Δy between the tunnels, and the stratum loss rate includes the stratum loss rates of the existing tunnel and the newly built tunnel, which are respectively expressed as and Specifically, Figure 5 As shown, the meaning of each tunnel variable in the present invention. Each working condition, i.e., the relative position of the tunnels is different, but other parameters are the same. In one embodiment of the present invention, the value range of the horizontal and vertical distances Δx and Δy between tunnels is set within 5D, and a group of working conditions is set every 0.5D.

[0102] The tunnel variables under different working conditions are input into the trained generalized regression neural network, that is, multiple sets of specific parameters are output, namely , , , , .

[0103] Step S20: Substitute all the specific parameters into the pre-constructed displacement equation to obtain the displacement equation of the existing tunnel.

[0104] Specifically, in the present invention, the corresponding displacement can be obtained by substituting the corresponding specific parameters into the pre-constructed displacement equation. The present invention uses a generalized regression neural network to predict the five key parameters, learns the patterns in the historical data through the neural network, predicts the undetermined parameters and substitutes them into the corresponding displacement formula for calculation, thereby achieving a complete prediction of the lining displacement curve.

[0105] Furthermore, substituting all the specific parameters into the pre-constructed displacement equation to obtain the displacement equation of the existing tunnel specifically includes:

[0106] Obtain the target demand and substitute all the specific parameters into the pre-built lining displacement equation;

[0107] The lining displacement equation after substituting the parameters is calculated to obtain the target equation, and the target equation is converted into the displacement equation according to the target requirements.

[0108] Specifically, in the present invention, all the specific parameters are substituted into the pre-constructed lining displacement equation, that is, into formula (1), and the lining displacement equation after the parameters are substituted is calculated to obtain the target equation. Then, according to the target requirements, it is determined which coordinate system the corresponding requirement is, and then it is converted into formula (2) or formula (3) to obtain the displacement equation.

[0109] Step S30: generating a lining displacement curve according to the displacement equation.

[0110] In the present invention, after the displacement equation is obtained, the corresponding displacement curve can be obtained on the corresponding coordinate system, and the displacement curve can be output.

[0111] Furthermore, if Figure 6As shown, the present invention trains a machine learning model through a data set, and then uses it for parameter prediction to obtain displacements that can be superimposed, thereby obtaining corresponding displacements under different working conditions. Compared with a simple machine learning model, the present invention combines the dual advantages of data-driven and knowledge-driven. The proposed method can obtain higher accuracy when predicting the lining displacement curve of an existing tunnel, and by decomposing the lining displacement into four parts: elliptical convergence, overall horizontal displacement, overall vertical displacement, and displacement deflection, it can better explain the specific morphological changes of the displacement curve. This decomposition method not only improves the understanding of the prediction results, but also helps analyze the influence mechanism of different factors on displacement during tunnel excavation, providing a scientific basis for engineering design and risk assessment; in addition, since the prediction method of the present invention can more accurately predict the displacement curve before the project is implemented, it reduces the uncertainty in the construction process, thereby effectively reducing the construction risk. This not only helps to prevent accidents during construction, but also optimizes the construction plan, reduces project delays and additional costs, and improves economic benefits.

[0112] The present invention obtains multiple tunnel variables between the existing tunnel and the newly built tunnel under target working conditions, inputs the multiple tunnel variables into a trained generalized regression neural network, and outputs multiple specific parameters, wherein training data is generated based on a two-dimensional finite element model, and a generalized regression neural network model is trained according to the training data; all the specific parameters are substituted into a pre-constructed displacement equation to obtain the displacement equation of the existing tunnel; and a lining displacement curve is generated according to the displacement equation. The present invention combines theoretical analysis with a neural network algorithm to decompose the lining displacement of the existing tunnel into four items and quantify it using five specific parameters, thereby deriving the displacement equation of each point on the lining, and substituting the parameters generated by the neural network into the corresponding displacement equation to obtain the corresponding displacement, thereby generating an accurate displacement curve.

[0113] Furthermore, if Figure 7 As shown, based on the above-mentioned cross-overlap tunnel excavation lining displacement curve prediction method, the present invention also provides a cross-overlap tunnel excavation lining displacement curve prediction system, wherein the cross-overlap tunnel excavation lining displacement curve prediction system includes:

[0114] A specific parameter output module 71 is used to obtain a plurality of tunnel variables between the existing tunnel and the newly built tunnel under target working conditions, input the plurality of tunnel variables into a trained generalized regression neural network, and output a plurality of specific parameters, wherein training data is generated based on a two-dimensional finite element model, and a generalized regression neural network model is trained according to the training data;

[0115] A displacement generating module 72, used for substituting all the specific parameters into a pre-constructed displacement equation to obtain the displacement equation of the existing tunnel;

[0116] The result output module 73 is used to generate a lining displacement curve according to the displacement equation.

[0117] Furthermore, if Figure 8 As shown, based on the above-mentioned method and system for predicting lining displacement curves in cross-overlapping tunnel excavation, the present invention also provides a terminal accordingly, and the terminal includes a processor 10, a memory 20 and a display 30. Figure 8 Only some components of the terminal are shown, but it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0118] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory of the terminal. In other embodiments, the memory 20 may also be an external storage device of the terminal, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the terminal. Further, the memory 20 may also include both an internal storage unit of the terminal and an external storage device. The memory 20 is used to store application software and various types of data installed on the terminal, such as the program code of the installation terminal. The memory 20 may also be used to temporarily store data that has been output or is to be output. In one embodiment, the memory 20 stores a cross-overlap tunnel excavation lining displacement curve prediction program 40, which can be executed by the processor 10, thereby realizing the cross-overlap tunnel excavation lining displacement curve prediction method of the present invention.

[0119] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor or other data processing chip, used to run the program code or process data stored in the memory 20, such as executing the cross-overlap tunnel excavation lining displacement curve prediction method.

[0120] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch device, etc. The display 30 is used to display information on the terminal and to display a visual user interface. The components 10-30 of the terminal communicate with each other via a system bus.

[0121] In one embodiment, when the processor 10 executes the cross-overlap tunnel excavation lining displacement curve prediction program 40 in the memory 20, the steps of the above cross-overlap tunnel excavation lining displacement curve prediction method are implemented.

[0122] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a cross-overlapping tunnel excavation lining displacement curve prediction program, and when the cross-overlapping tunnel excavation lining displacement curve prediction program is executed by a processor, the steps of the cross-overlapping tunnel excavation lining displacement curve prediction method as described above are implemented.

[0123] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the existence of other identical elements in the process, method, article or terminal including the element.

[0124] Of course, those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing related hardware (such as a processor, a controller, etc.) through a computer program, and the program can be stored in a computer-readable storage medium that can be read by a computer, and the program can include the processes of the above-mentioned method embodiments when executed. The computer-readable storage medium can be a memory, a disk, an optical disk, etc.

[0125] It should be understood that the application of the present invention is not limited to the above examples. For ordinary technicians in this field, improvements or changes can be made based on the above description. All these improvements and changes should fall within the scope of protection of the claims attached to the present invention.

Claims

1. A method for predicting lining displacement curves for overlapping tunnel excavation, characterized in that: The method for predicting displacement curves of linings in cross-overlap tunnel excavation includes: Acquire multiple tunnel variables between the existing tunnel and the newly built tunnel under target working conditions, input the multiple tunnel variables into a trained generalized regression neural network, and output multiple specific parameters, wherein training data is generated based on a two-dimensional finite element model, and the generalized regression neural network model is trained according to the training data; Substituting all the specific parameters into a pre-constructed displacement equation to obtain the displacement equation of the existing tunnel; generating a lining displacement curve according to the displacement equation; The step of obtaining a plurality of tunnel variables between the existing tunnel and the newly built tunnel under target working conditions, inputting the plurality of tunnel variables into a trained generalized regression neural network, and outputting a plurality of specific parameters specifically includes: Acquire multiple tunnel variables between the existing tunnel and the newly built tunnel under target working conditions, wherein the tunnel variables include tunnel diameter, tunnel burial depth, stratum loss rate and tunnel relative position, and the tunnel relative position in the tunnel variables under different working conditions is different; The plurality of tunnel variables under the target working condition are input into the trained generalized regression neural network, and a plurality of specific parameters are output, wherein each group of specific parameters includes a first elliptical convergence, a second elliptical convergence, an overall horizontal displacement, an overall vertical displacement and an overall displacement deflection.

2. The method for predicting displacement curves of linings in cross-overlap tunnel excavation according to claim 1 is characterized in that: The training process of the generalized regression neural network specifically includes: Generating the training data based on the two-dimensional finite element model, and dividing the training data into a training set and a validation set; The generalized regression neural network is trained according to the training set, wherein the kernel width is optimized according to the grid search method and the cross-validation method in each training, and the corresponding determination coefficient is calculated after each training, and when the determination coefficient reaches the preset requirement, the training is terminated to obtain a preliminarily trained generalized regression neural network; Inputting the verification set into a preliminarily trained generalized regression neural network, outputting a verification result, generating a first verification displacement curve according to the verification result, and generating a comparison result according to the verification set and the first verification displacement curve; When the comparison result meets the preset comparison requirement, the current generalized regression neural network is used as the trained generalized regression neural network.

3. The method for predicting displacement curves of linings in cross-overlap tunnel excavation according to claim 2 is characterized in that: Generating the training data based on the two-dimensional finite element model and dividing the training data into a training set and a validation set specifically includes: Acquire training tunnel variables, wherein the training tunnel variables include training tunnel diameter, training tunnel burial depth, training stratum loss rate and tunnel relative position; constructing a two-dimensional tunnel model in the two-dimensional finite element model based on a training tunnel project and the training tunnel variables; The tunnel construction process is simulated in the two-dimensional finite element model to obtain the training specific parameters corresponding to the training tunnel variables, the training data is constructed according to the training specific parameters corresponding to the training tunnel variables, and the training data is divided into a training set and a validation set.

4. The method for predicting displacement curves of linings in cross-overlap tunnel excavation according to claim 2 is characterized in that: The step of inputting the verification set into a preliminarily trained generalized regression neural network, outputting a verification result, generating a first verification displacement curve according to the verification result, and generating a comparison result according to the verification set and the first verification displacement curve specifically includes: Inputting the verification set into a preliminarily trained generalized regression neural network and outputting a verification result; Substituting the verification result into a pre-constructed displacement equation to generate a plurality of training displacements, generating a plurality of first verification displacement curves according to the plurality of training displacement equations, and generating a plurality of second verification displacement curves according to the training specific parameters corresponding to each training tunnel variable in the verification set; A comparison result is generated according to all of the first verification displacement curves and the corresponding second verification displacement curves.

5. The method for predicting displacement curves of linings in cross-overlap tunnel excavation according to claim 1 is characterized in that: Substituting all the specific parameters into the pre-constructed displacement equation to obtain the lining displacement equation of the existing tunnel also includes: Presetting the parameter representation of the first elliptical convergence, the second elliptical convergence, the overall horizontal displacement, the overall vertical displacement and the overall displacement deflection; The lining displacement equations are generated based on the parametric representation of the first elliptical convergence, the second elliptical convergence, the overall horizontal displacement, the overall vertical displacement and the overall displacement deflection.

6. The method for predicting displacement curves of linings in cross-overlap tunnel excavation according to claim 1 is characterized in that: Substituting all the specific parameters into the pre-constructed displacement equation to obtain the displacement equation of the existing tunnel specifically includes: Obtain the target demand and substitute all the specific parameters into the pre-built lining displacement equation; The lining displacement equation after substituting the parameters is calculated to obtain the target equation, and the target equation is converted into the displacement equation according to the target requirements.

Images