Tunnel excavation deformation characteristic curve prediction method based on proxy model

Abstract

The invention discloses a tunnel excavation deformation characteristic curve prediction method based on an agent model. The method comprises the following steps: step 1, determining an input variable and a value range thereof; 2, parameterizing and representing a tunnel excavation deformation characteristic curve; step 3, establishing an agent model and training; 4, tunnel excavation deformation characteristic curve parameters are predicted; and 5, drawing a tunnel excavation deformation characteristic curve. According to the method, through parameterization characterization of the tunnel excavation deformation characteristic curve and establishment of the tunnel excavation deformation characteristic curve parameter prediction agent model, corresponding tunnel excavation deformation characteristic curve parameter values are directly obtained under the condition that finite element modeling analysis does not need to be carried out in the tunnel excavation construction process; therefore, the tunnel excavation deformation characteristic curve is quickly predicted, the prediction workload is reduced, and the prediction efficiency is remarkably improved. The bottleneck that a traditional simulation analysis prediction method cannot rapidly predict the tunnel excavation deformation characteristic curve is broken through, and the method has wide application prospects in tunnel construction safety prevention and control and support optimization design.

Description

Technical Field

[0001] The invention relates to the field of tunnel structure morphology prediction and analysis, in particular to a method for predicting tunnel excavation deformation characteristic curve based on an agent model. Background Art

[0002] Hydraulic tunnels are an important part of the water conservancy and hydropower infrastructure system, which need to be excavated and constructed in the mountains or underground. Affected by the disturbance of tunnel excavation, the surrounding rock continues to deform during the construction process and may become deformed and unstable. Therefore, the reasonable prediction of tunnel excavation deformation characteristics is the premise and basis for carrying out tunnel construction safety prevention and control and optimizing the design of excavation support schemes. The deformation characteristics of tunnel excavation are affected by multiple factors such as burial depth, tunnel diameter, cycle footage and surrounding rock mechanical parameters. At the same time, they involve variable structural nonlinearity (nonlinearity caused by changes in surrounding rock structure caused by tunnel excavation) and material nonlinearity (nonlinearity caused by nonlinear stress-strain relationship of surrounding rock), which are complex and changeable. Since the deformation of tunnel surrounding rock gradually decreases from the excavation boundary to the deep part of surrounding rock, the focus of tunnel design and construction is on the change process of radial displacement of tunnel vertex with the distance between predicted section and excavation surface, and it is described by tunnel excavation deformation characteristic curve.

[0003] At present, the prediction of tunnel excavation deformation characteristic curve mainly adopts simulation analysis prediction method, that is, by using finite element and other numerical analysis methods to simulate the tunnel excavation construction process to predict the distribution, value and development and evolution of surrounding rock deformation during the advancement of the excavation working face, so as to draw a curve of the change of the tunnel vertex at the predicted section with the distance between the predicted section and the excavation face. The above-mentioned existing technology has the following deficiencies:

[0004] 1. In the existing technology, predicting the deformation characteristic curve of tunnel excavation requires the establishment of a complex finite element simulation model and the implementation of corresponding calculation and analysis, which is labor-intensive, time-consuming, and has low prediction efficiency.

[0005] 2. In the prior art, when the tunnel design and construction scheme or the surrounding rock mechanical parameters are adjusted, the corresponding finite element simulation modeling analysis needs to be re-conducted, and it is impossible to quickly predict the tunnel excavation deformation characteristic curve under different design and construction schemes or parameter conditions. Summary of the invention

[0006] Purpose of the invention: In order to overcome the deficiencies of the above-mentioned prior art, the present invention provides a method for predicting the deformation characteristic curve of tunnel excavation based on a proxy model. This method breaks through the bottleneck that traditional simulation analysis and prediction methods cannot quickly predict the deformation characteristic curve of tunnel excavation by parameterizing the characterization of the deformation characteristic curve of tunnel excavation and establishing a parameter prediction proxy model of the deformation characteristic curve of tunnel excavation, thereby greatly improving the prediction efficiency. This method has broad application prospects in tunnel construction safety prevention and control and support optimization design.

[0007] To solve the technical problem raised in the background technology: the present invention provides a method for predicting tunnel excavation deformation characteristic curve based on a proxy model, comprising the following steps:

[0008] Step 1: Determine the input variables: Select the factors directly related to the tunnel stress as the main influencing factors among the factors affecting the deformation characteristics of tunnel excavation. The main influencing factors include tunnel burial depth H, tunnel diameter D, excavation cycle footage f p , deformation modulus E of surrounding rock, Poisson's ratio μ, internal friction angle φ and cohesion c;

[0009] Among the main factors affecting the deformation characteristics of tunnel excavation, N p Input variable P i , and determine the value range of each input variable [l i ,m i ];

[0010] Step 2: Parameterize and characterize the tunnel excavation deformation characteristic curve: A three-parameter exponential function is used to characterize the tunnel excavation deformation characteristic curve. The function expression is as follows:

[0011]

[0012] Where u is the radial displacement of the tunnel vertex at the predicted section; L is the distance between the predicted section and the excavation surface; u max is the parameter representing the maximum value of excavation deformation; k is the parameter representing the rate of change of deformation rate of tunnel excavation deformation characteristic curve; a is the parameter representing the excavation deformation and u when L = 0 max Parameters of the proportional relationship between

[0013] Step 3, establish and train the proxy model: construct a training sample set S, train the proxy model through the model input and model output of each sample in the sample set S, and obtain the tunnel excavation deformation characteristic curve parameter prediction proxy model;

[0014] Step 4: Parameter prediction: Use the proxy model trained in step 3 to predict the given input variable value P i * (i=1,2,…,N p ) is the model input to predict the tunnel excavation deformation characteristic curve parameters;

[0015] Step 5: Result drawing: Draw the tunnel excavation deformation characteristic curve according to the parameters predicted in step 4.

[0016] Furthermore, the P i , 1≤i≤N p .

[0017] Furthermore, in step 3, the steps of establishing and training the proxy model include:

[0018] Step 31: extract N s Sample S j ,j∈[1,N s ], and obtain the model input of each sample in the proxy model training sample set S, where the jth sample S j The model input is denoted as P i j ;

[0019] Step 32: N s Sample S j (j=1,2,…,N s ) is used as the model input, and the corresponding tunnel excavation finite element modeling calculation is carried out to establish a finite element simulation model of the tunnel excavation process, and simulate the tunnel excavation process. j The model input P i j The corresponding tunnel excavation finite element model is denoted as M j ;

[0020] Step 33: Substitute the calculation results of each tunnel excavation finite element modeling into the three-parameter exponential function for fitting, and obtain the tunnel excavation deformation characteristic curve characterization parameter u max A set of fitted values ​​of , k and a, based on the tunnel excavation finite element model M j The characterization parameter u obtained by fitting the calculation results max The fitted values ​​of , k and a are recorded as u maxj , k j and a j , with u maxj , k j and a j As the basis for construction and the j-th sample S j Model output in Equal to u maxj , k j 、a j ;

[0021] Step 34: Use the sample set S to train the selected proxy model to obtain a tunnel excavation deformation characteristic curve parameter prediction proxy model.

[0022] Further, in step 31, the input space of the tunnel excavation deformation characteristic curve parameter prediction proxy model is composed of P i N p dimensional bounded space, and P i ∈[l i ,mi ].

[0023] Furthermore, in step 32, the finite element simulation model of the tunnel excavation process uses the predicted section as the middle cross-section of the model, with an axial length ≥15 times the tunnel diameter, the center of the cross-section located at the center of the tunnel, and the vertical distance between the center of the cross-section and the surrounding boundaries ≥5.5 times the tunnel diameter.

[0024] Furthermore, in step 4, the input variable value P i * where i∈[1,N p ], and input variable value P i * With input variable P i The multiple influencing factors correspond one to one.

[0025] Furthermore, the process of predicting the parameters of tunnel excavation deformation characteristic curve is as follows:

[0026] Enter the given input variable value The established tunnel excavation deformation characteristic curve parameter prediction proxy model is used to obtain the corresponding model output With P i * The predicted value u of the corresponding tunnel excavation deformation characteristic curve parameter max * , k * 、a * Equal to

[0027] Furthermore, in step 5, the specific process of drawing the tunnel excavation deformation characteristic curve is as follows: let u in the three-parameter exponential function be max , k and a take the corresponding predicted value u max * , k * and a * , draw the function curve according to the function expression after determining the parameters.

[0028] The present invention has the following beneficial effects:

[0029] The present invention parameterizes the tunnel excavation deformation characteristic curve and establishes a tunnel excavation deformation characteristic curve parameter prediction proxy model. After the tunnel excavation deformation characteristic curve parameter prediction proxy model is established, for any given tunnel excavation deformation characteristic curve prediction input variable, the corresponding tunnel excavation deformation characteristic curve parameter value is directly obtained without the need to carry out finite element modeling analysis of the tunnel excavation construction process, thereby quickly predicting the tunnel excavation deformation characteristic curve, reducing the prediction workload, reducing work time, and significantly improving the prediction efficiency. In the actual working process, even if the tunnel design and construction plan or the surrounding rock mechanical parameters change, the proxy model in the method is still applicable, and there is no need to re-carry out the corresponding finite element simulation modeling analysis. This solution breaks through the bottleneck that the prior art cannot quickly predict the tunnel excavation deformation characteristic curve, and provides an effective solution for efficiently carrying out tunnel construction safety prevention and control and support optimization design. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 The flowchart of the tunnel excavation deformation characteristic curve prediction method based on the agent model.

[0031] Figure 2 This is a finite element simulation model diagram of the tunnel excavation process in the embodiment provided by the present invention.

[0032] Figure 3 An example diagram of parameter fitting of a tunnel excavation deformation characteristic curve in an embodiment provided by the present invention.

[0033] Figure 4 The tunnel excavation deformation characteristic curve predicted in the embodiment provided by the present invention. DETAILED DESCRIPTION

[0034] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0035] like Figure 1 As shown in FIG. 1 , the tunnel excavation deformation characteristic curve prediction method based on the agent model includes the following steps:

[0036] Step 1: Determine the input variables and their value ranges for predicting the tunnel excavation deformation characteristic curve. The specific process is as follows:

[0037] Among the factors affecting the deformation characteristics of tunnel excavation, factors directly related to the tunnel stress are selected as the main influencing factors. The main influencing factors include tunnel burial depth H, tunnel diameter D, excavation cycle footage f p , surrounding rock deformation modulus E, Poisson's ratio μ, internal friction angle φ and cohesion c; collect N from the main influencing factors of tunnel excavation deformation characteristics p factors, as N for predicting tunnel excavation deformation characteristic curve p Input variable Pi (i=1,2,…,N p ), and determine the value range of each input variable. The i-th input variable P i The value range of is denoted as [l i ,m i ].

[0038] The tunnel excavation deformation characteristic curve refers to the relationship curve between the radial displacement u of the tunnel vertex at the predicted section and the distance L between the predicted section and the excavation surface.

[0039] In this embodiment, a water diversion tunnel is about 3.5 km long and has a burial depth of about 260-460 m. The rock strata along the line are steeply inclined and the rock properties are complex and diverse. Different tunnel sections have different excavation deformation characteristics. The tunnel cross-section is circular, the excavation diameter is 6.8 m, and the excavation cycle footage is 2 m.

[0040] According to the actual situation of the project, considering the tunnel diameter D along the tunnel and the excavation cycle footage f p Keeping the surrounding rock deformation modulus E, Poisson's ratio μ, internal friction angle φ, cohesion c and burial depth H unchanged, the five factors of surrounding rock deformation modulus E, Poisson's ratio μ, internal friction angle φ, cohesion c and burial depth H are selected as the five input variables for predicting the tunnel excavation deformation characteristic curve, and the surrounding rock deformation modulus E, Poisson's ratio μ, internal friction angle φ, cohesion c and burial depth H are denoted as P 1 , P 2 , P 3 , P 4 and P 5 , based on the engineering geological survey data, determine the reasonable value range of each input variable, as shown in Table 1.

[0041] Table 1 Prediction input variables and their value ranges of tunnel excavation deformation characteristic curve

[0042] <![CDATA[P 1 (GPa)]]> <![CDATA[P 2 ]]> <![CDATA[P 3 (°)]]> <![CDATA[P 4 (MPa)]]> <![CDATA[P 5 (m)]]> [6,10] [0.24,0.28] [35,45] [0.5,1] [260,460]

[0043] Step 2: Parameterize and characterize the tunnel excavation deformation characteristic curve: A three-parameter exponential function is used to characterize the tunnel excavation deformation characteristic curve. The function expression is as follows:

[0044]

[0045] Where L is the distance between the predicted section and the excavation surface; u is the radial displacement of the tunnel vertex at the predicted section; u max is the parameter representing the maximum value of excavation deformation; k is the parameter representing the rate of change of deformation rate of tunnel excavation deformation characteristic curve; a is the parameter representing the excavation deformation and u when L = 0 max Parameters of the proportional relationship between them.

[0046] Step 3: Establish and train the proxy model: construct a training sample set S, train the proxy model through the model input and model output of each sample in the sample set S, and obtain the tunnel excavation deformation characteristic curve parameter prediction proxy model. The steps include:

[0047] Step 31: Sampling in the tunnel excavation deformation characteristic curve parameter prediction proxy model input space to obtain N of the proxy model training sample set S s Sample S j (j=1,2,…,N s ) model input, where the jth sample S j The model input is denoted as P i j .

[0048] The input space of the tunnel excavation deformation characteristic curve parameter prediction proxy model is composed of P i (i=1,2,…,N p ) p dimensional bounded space, and P i ∈[l i ,m i ].

[0049] In this embodiment, the input space of the tunnel excavation deformation characteristic curve parameter prediction agent model is composed of P i (i=1,2,…,5) constitutes a 5-dimensional bounded space, P 1 ∈[6,10], P 2 ∈[0.24,0.28], P 3 ∈[35,45], P 4 ∈[0.5,1], P 5 ∈[260,460]; Latin hypercube sampling technology is used to sample in the input space of the proxy model for predicting tunnel excavation deformation characteristic curve parameters. The number of samples included in the proxy model training sample set S is taken as 30. The 30 samples S j (j=1,2,…,30) model input, as shown in Table 2, the jth sample S j The model input is denoted as P i j (i=1,2,…,5; j=1,2,…,30).

[0050] Table 2 Model input of each sample in the proxy model training sample set S

[0051]

[0052]

[0053] Step 32: For each sample model input in the training sample set S, the corresponding tunnel excavation finite element modeling calculation is carried out, that is, N s Sample S j (j=1,2,…,N s ) is used as the modeling parameter, and the corresponding tunnel excavation finite element modeling calculation is carried out. j The model input P i j The corresponding tunnel excavation finite element model is denoted as M j .

[0054] Finite element modeling calculation for tunnel excavation refers to establishing a finite element simulation model of the tunnel excavation process and simulating the tunnel excavation process by taking the model input of each sample in the training sample set S as the modeling parameter.

[0055] The finite element simulation model of the tunnel excavation process uses the predicted section as the middle cross-section of the model, with an axial length of not less than 15 times the tunnel diameter, the center of the cross-section located in the center of the tunnel, and the vertical distance between the center of the cross-section and the surrounding boundaries not less than 5.5 times the tunnel diameter.

[0056] In this embodiment, the axial length of the finite element simulation model of the tunnel excavation process is taken as 15 times the tunnel diameter (102m), and the vertical distance between the cross-section center and the surrounding boundaries is taken as 5.5 times the tunnel diameter (37.4m). Figure 2 shown.

[0057] Step 33: Based on the calculation results of the finite element modeling of each tunnel excavation, multiple calculation results are sequentially substituted into the three-parameter exponential function for fitting to obtain the tunnel excavation deformation characteristic curve characterization parameter u max A set of fitted values ​​of , k and a, based on the tunnel excavation finite element model M j The characterization parameter u obtained by fitting the calculation results max The fitted values ​​of , k and a are recorded as u maxj , k j and a j , with u maxj , k j and a j As the basis for construction and the j-th sample S j Model output (m=1,2,3), where Q 1 j , Equal to u maxj , k j 、a j .

[0058] The calculation results of tunnel excavation finite element modeling refer to the change process of the radial displacement u of the tunnel vertex at the predicted section obtained by simulating the tunnel excavation process as the distance L between the predicted section and the excavation surface changes.

[0059] In this embodiment, based on the finite element modeling calculation results of 30 tunnel excavations, a three-parameter exponential function is used as the fitting function to fit the change process of the radial displacement u of the tunnel vertex at the predicted section as the distance L between the predicted section and the excavation surface changes ( Figure 3 ), and 30 sets of tunnel excavation deformation characteristic curve characterization parameters u are obtained max , k and the fitted values ​​u of a maxj , k j and a j , j = 1, 2, ..., 30, let Q 1 j =u maxj , The model output of 30 samples in the proxy model training sample set S can be constructed (m=1,2,3;j=1,2,…,30), as shown in Table 3.

[0060] Table 3 Model output of each sample in the proxy model training sample set S

[0061]

[0062]

[0063] Step 34: Use the sample set S to train the selected proxy model to obtain a tunnel excavation deformation characteristic curve parameter prediction proxy model.

[0064] In this embodiment, the Kriging proxy model is selected as the proxy model for predicting the tunnel excavation deformation characteristic curve parameters, and 30 samples S in the training sample set S are used. j (j=1,2,…,30) model input (Table 2) and model output (Table 3) are used to train it. enter, As the model output of the prediction agent model, the prediction agent model of tunnel excavation deformation characteristic curve parameters can be obtained. The specific expression is:

[0065]

[0066] Among them, r is the 30th order correlation vector, which is determined by the predicted input variable P i * The model input P of the training sample i j The Gaussian correlation function values ​​between them are composed; R is the 30th order correlation matrix, which is composed of the model input P of the training sample ij The Gaussian correlation function between them is composed; F is a 30-order unit array; β 0 is a scalar, representing the constant term of the trained model output, and β 0 =(F T R -1 F) -1 F T R - 1 Y.

[0067] Step 4: Parameter prediction: Use the proxy model trained in step 3 to predict the given input variable value P i * (i=1,2,…,N p ) is the model input to predict the tunnel excavation deformation characteristic curve parameters. The specific process is as follows:

[0068] Enter the given input variable value P i * (i=1,2,…,N p ),Right now The established tunnel excavation deformation characteristic curve parameter prediction proxy model is used to obtain the corresponding model output With P i * The predicted value u of the corresponding tunnel excavation deformation characteristic curve parameter max * , k * 、a * Equal to Q 1 * , Q 2 * , Q 3 * .

[0069] In this embodiment, three typical tunnel sections (1 # , 2 # and 3 # tunnel section), of which 1 # , 2 # and 3 # The design and construction schemes of the tunnel sections (including the burial depth H) are different, and the mechanical parameters of the surrounding rock (including deformation modulus E, Poisson’s ratio μ, internal friction angle φ, and cohesion c) are also different. The established tunnel excavation deformation characteristic curve parameter prediction agent model is used to predict the excavation deformation characteristic curve parameters. The excavation deformation characteristic prediction input variables of each typical tunnel section determined based on the tunnel topographic and geological data are shown in Table 4, and the predicted values ​​of the excavation deformation characteristic parameters are shown in Table 5.

[0070] Table 4 Input variables for prediction of deformation characteristics of typical tunnel sections

[0071]

[0072] Table 5 Prediction values ​​of deformation characteristic curve parameters for typical tunnel sections

[0073]

[0074] Step 5: Result drawing: Draw the tunnel excavation deformation characteristic curve according to the parameters predicted in step 4. The specific process is as follows:

[0075] Let u in the three-parameter exponential function be max , k and a are equal to the predicted values ​​u of the tunnel excavation deformation characteristic curve parameters max * , k * and a * , draw the function curve, which is the predicted tunnel excavation deformation characteristic curve, such as Figure 4 shown.

[0076] The above embodiments are only preferred embodiments of the present invention and are not limitations of the technical solutions of the present invention. Any technical solution that can be implemented on the basis of the above embodiments without creative work should be deemed to fall within the scope of protection of the patent of the present invention.

Claims

1. A method for predicting tunnel excavation deformation characteristic curve based on a proxy model, characterized in that: The following steps are involved: Step 1: Determine the input variables: Select the factors directly related to the tunnel stress as the main influencing factors among the factors affecting the deformation characteristics of tunnel excavation. The main influencing factors include tunnel burial depth H, tunnel diameter D, excavation cycle footage f p , deformation modulus E of surrounding rock, Poisson's ratio μ, internal friction angle φ and cohesion c; Among the main factors affecting the deformation characteristics of tunnel excavation, N p Input variable P i , and determine the value range of each input variable [l i ,m i ]; Step 2: Parameterize and characterize the tunnel excavation deformation characteristic curve: A three-parameter exponential function is used to characterize the tunnel excavation deformation characteristic curve. The function expression is as follows: Where u is the radial displacement of the tunnel vertex at the predicted section; L is the distance between the predicted section and the excavation surface; u max is the parameter representing the maximum value of excavation deformation; k is the parameter representing the rate of change of deformation rate of tunnel excavation deformation characteristic curve; a is the parameter representing the excavation deformation and u when L = 0 max Parameters of the proportional relationship between Step 3, establish and train the proxy model: construct a training sample set S, train the proxy model through the model input and model output of each sample in the sample set S, and obtain the tunnel excavation deformation characteristic curve parameter prediction proxy model; Step 4: Parameter prediction: Use the proxy model trained in step 3 to predict the given input variable value P i * As model input, predict the parameters of tunnel excavation deformation characteristic curve; Step 5: Result drawing: Draw the tunnel excavation deformation characteristic curve according to the parameters predicted in step 4.

2. The method for predicting tunnel excavation deformation characteristic curve based on agent model according to claim 1, characterized in that: The P i , 1≤i≤N p .

3. The method for predicting tunnel excavation deformation characteristic curve based on agent model according to claim 1, characterized in that: In step 3, the steps of building and training the proxy model include: Step 31: extract N s Sample S j ,j∈[1,N s ], and obtain the model input of each sample in the proxy model training sample set S, where the jth sample S j The model input is denoted as P i j ; Step 32: N s Sample S j The model input is used as the modeling parameter, and the corresponding tunnel excavation finite element modeling calculation is carried out to establish a finite element simulation model of the tunnel excavation process and simulate the tunnel excavation process. j The model input P i j The corresponding tunnel excavation finite element model is denoted as M j ; Step 33: Substitute the calculation results of each tunnel excavation finite element modeling into the three-parameter exponential function for fitting, and obtain the tunnel excavation deformation characteristic curve characterization parameter u max A set of fitted values ​​of , k and a, based on the tunnel excavation finite element model M j The characterization parameter u obtained by fitting the calculation results max The fitted values ​​of , k and a are recorded as u maxj , k j and a j , with u maxj , k j and a j As the basis for construction and the j-th sample S j Model output m∈[1,3], where are equal to u maxj , k j 、a j ; Step 34: Use the sample set S to train the selected proxy model to obtain a tunnel excavation deformation characteristic curve parameter prediction proxy model.

4. The method for predicting tunnel excavation deformation characteristic curve based on agent model according to claim 3 is characterized in that: In step 31, the input space of the tunnel excavation deformation characteristic curve parameter prediction proxy model is composed of P i N p dimensional bounded space, and P i ∈[l i ,m i ].

5. The method for predicting tunnel excavation deformation characteristic curve based on agent model according to claim 3, characterized in that: In step 32, the finite element simulation model of the tunnel excavation process uses the predicted section as the middle cross-section of the model, with an axial length ≥15 times the tunnel diameter, the center of the cross-section located at the center of the tunnel, and the vertical distance between the center of the cross-section and the surrounding boundaries ≥5.5 times the tunnel diameter.

6. The method for predicting tunnel excavation deformation characteristic curve based on agent model according to claim 3, characterized in that: In step 4, the input variable value P i * where i∈[1,N p ], and input variable value P i * With input variable P i The multiple influencing factors correspond one to one.

7. The method for predicting tunnel excavation deformation characteristic curve based on agent model according to claim 6, characterized in that: The process of predicting the parameters of tunnel excavation deformation characteristic curve is as follows: Enter the given input variable value P1 * , The established tunnel excavation deformation characteristic curve parameter prediction proxy model is used to obtain the corresponding model output With P i * The predicted value u of the corresponding tunnel excavation deformation characteristic curve parameter max * , k * 、a * Equal to 8. The method for predicting tunnel excavation deformation characteristic curve based on agent model according to claim 7, characterized in that: In step 5, the specific process of drawing the tunnel excavation deformation characteristic curve is as follows: let u in the three-parameter exponential function max , k and a take the corresponding predicted value u max * , k * and a * , draw the function curve according to the function expression after determining the parameters.

Images