A method and device for simulating the impact of tunnels on a landslide section under vibration load

Abstract

The present invention discloses a tunnel impact simulation test method and device for a landslide section under vibration load, which belongs to the field of tunnel simulation test for landslide sections, and includes the following steps: S1, constructing a multi-hazard damage prediction model to obtain tunnel landslide damage prediction data; S2, collecting tunnel landslide damage test data; S3, calculating the degree of deviation between the tunnel landslide damage prediction data and the tunnel landslide damage test data, and judging whether the degree of deviation is greater than a threshold, if so, adjusting the boundary conditions and returning to step S2; if not, outputting the corresponding relationship between the current boundary conditions and the tunnel landslide parameters to obtain a corresponding relationship table; S4, collecting tunnel landslide damage data and outputting it for display. The above-mentioned tunnel impact simulation test method and device for a landslide section under vibration load can provide more accurate and reliable prediction and test results in the tunnel impact simulation test for a landslide section under vibration load, and provide strong support for engineering design and construction.

Description

Technical Field

[0001] The invention relates to the technical field of landslide tunnel simulation test, and in particular to a landslide tunnel impact simulation test method and device under vibration load. Background Art

[0002] The two most common natural disasters that cause major structural damage are earthquakes and landslides. For mountainous areas, in addition to ground rupture and structural deformation caused by earthquakes, landslide disasters can also occur. During tunnel construction and route selection, landslides are often crossed, forming a tunnel-landslide system. The relative position between the tunnel and the sliding surface determines the deformation characteristics of the sliding surface and the stress pattern of the tunnel. Earthquakes, as the main factor causing landslide deformation and tunnel damage, are uncertain and accidental, causing landslides to gradually deform and lose stability under earthquake loads, and then the tunnel structure will be affected by the landslide, which has a significant impact on the stability of the tunnel-landslide system.

[0003] In order to study the impact of earthquakes on tunnels in landslide areas, the existing technology generally uses a shaking table simulation test, and the test steps are as follows: 1. Model installation: Fix the tunnel model and landslide model on the shaking table to ensure the stability of the model; install the sensor at the same time to ensure that the sensor position is accurate and the connection is reliable. 2. Preliminary experiment: Perform preliminary loading and check the operating status of the sensor and data acquisition system. And adjust the sensor position and parameters to ensure the accuracy of data acquisition. 3. Formal experiment: Loading seismic motion: According to the predetermined seismic motion parameters, apply seismic motion through the shaking table. In this process, the acceleration, displacement, strain and other data are recorded in real time, and the response phenomena of the tunnel and landslide models, such as cracks, displacement, deformation, etc., are recorded at the same time. 4. Multiple experiments: Change parameters: Adjust the seismic motion parameters (such as frequency, amplitude, duration), conduct multiple experiments, compare the experimental results under different parameters, and analyze the influence of seismic motion on the tunnel and landslide models.

[0004] It can be seen that the existing tests have the following defects:

[0005] 1. Existing boundary conditions are mostly based on experience and lack scientific basis, resulting in large deviations between test results and actual results;

[0006] 2. Due to the limitations of laboratory space and equipment, experiments usually use small models, which may lead to scale effects, that is, the response of the model is different from the response in the actual project. Summary of the invention

[0007] The purpose of the present invention is to provide a method and device for simulating the impact of tunnels on a landslide section under vibration load, so as to solve the above technical problems.

[0008] To achieve the above object, the present invention provides a vibration load tunnel impact simulation test method in a landslide section, comprising the following steps:

[0009] S1. Construct a multi-hazard damage prediction model, and input historical earthquake parameters and tunnel landslide parameters into the multi-hazard damage prediction model to obtain tunnel landslide damage prediction data;

[0010] S2, based on the same tunnel landslide parameters as step S1, construct a tunnel landslide test model, then place the tunnel landslide test model on a shaking table, and then based on the initial boundary conditions, input the same earthquake parameters as step S1 into the shaking table to collect tunnel landslide damage test data;

[0011] S3, calculating the deviation between the tunnel landslide damage prediction data obtained in step S1 and the tunnel landslide damage test data collected in step S2, and judging whether the deviation is greater than a threshold value, if so, adjusting the boundary conditions and returning to step S2; if not, outputting the corresponding relationship between the current boundary conditions and the tunnel landslide parameters to obtain a corresponding relationship table;

[0012] S4. Obtain tunnel landslide test parameters according to similarity theory, and construct a tunnel landslide test model based on the tunnel landslide test parameters. At the same time, based on the tunnel landslide test parameters, search the corresponding relationship table described in step S3 to obtain boundary conditions. Based on the obtained boundary conditions, input earthquake parameters to the shaking table, collect tunnel landslide damage data and output them for display.

[0013] Preferably, step S1 specifically includes the following steps:

[0014] S11, using the collected historical earthquake parameters and tunnel landslide parameters to form a data set, and dividing the data set into a training set and a test set according to the ratio;

[0015] Earthquake parameters include earthquake acceleration ,earthquake magnitude , epicenter distance and earthquake duration ; Tunnel landslide parameters include landslide hazard index , Geological Strength Index and damage index ; Among them, the earthquake acceleration ,earthquake magnitude , epicenter distance , Earthquake duration , Landslide Hazard Index and geological intensity index are the input vectors of the multi-hazard damage prediction model , damage index is the output vector of the multi-hazard damage prediction model ;

[0016] S12. A multi-hazard damage prediction model is constructed using a feedforward neural network algorithm, wherein the multi-hazard damage prediction model is expressed as follows:

[0017] ;

[0018] In the formula, represents the distance coefficient; represents the terrain factor; An index indicating the magnitude of an earthquake; Represents earthquake acceleration and earthquake magnitude Index of Represents terrain factor Index of

[0019] S13, inputting the training set into the multi-hazard damage prediction model, training the multi-hazard damage prediction model using the Levenberg-Marquardt back propagation algorithm and the gradient descent weight and bias learning function with momentum, repeatedly assigning the weight of each neuron through the training set to establish a correlation between the input and output layers of the multi-hazard damage prediction model;

[0020] S14, inputting the test set into the multi-hazard damage prediction model to test the performance of the multi-hazard damage prediction model;

[0021] S15. Evaluate the performance of the multi-hazard damage prediction model using performance evaluation indicators.

[0022] Preferably, the multi-hazard damage prediction model described in step S12 includes an input layer, three hidden layers and an output layer;

[0023] The logarithmic sigmoid transfer function is used as the nonlinear activation function of the hidden layer neurons, and the linear transfer function is used as the transfer function of the output layer.

[0024] Preferably, in step S12, Take 0.003, Take 3.9, Take 5.3.

[0025] Preferably, step S13 specifically includes the following steps:

[0026] S131, initializing a multi-hazard damage prediction model;

[0027] S132, forward propagation: input the training set into the multi-hazard damage prediction model and calculate the output of each layer;

[0028] The output expression from the input layer to the hidden layer is as follows:

[0029] ;

[0030] ;

[0031] In the formula, Indicates The net input to the hidden layer neurons; Represents the first neurons to the hidden layer The connection weights of neurons; Represents the input layer The input features of each neuron; represents the hidden layer The bias of each neuron; represents the hidden layer The output of a neuron; represents the number of neurons in the input layer; represents the activation function;

[0032] The output expression from the hidden layer to the output layer is as follows:

[0033] ;

[0034] ;

[0035] In the formula, Indicates The net input to the output layer neurons; represents the number of neurons in the hidden layer; Indicates that from the hidden layer The first neuron to the output layer The connection weights of neurons; The output layer The bias of each neuron; The output layer The output of each neuron is considered as the network prediction output;

[0036] S133, calculate the target output of each sample in the training set and the network prediction output The difference between :

[0037] ;

[0038] S134. Back propagation: computing differences Gradients with respect to weights;

[0039] Among them, the difference from the output layer to the hidden layer The formula for calculating the gradient of weight is as follows:

[0040] ;

[0041] ;

[0042] ;

[0043] Differences between hidden layer and input layer The formula for calculating the gradient of weight is as follows:

[0044] ;

[0045] ;

[0046] ;

[0047] In the formula, Represents the output layer The error term of each neuron; Represents the activation function exist The derivative at ; Indicates that from the hidden layer neurons to the output layer The update amount of the weight of each neuron; represents the learning rate; represents the momentum term coefficient; Indicates that the last iteration from the hidden layer neurons to the output layer The weight update amount of each neuron; Represents the output layer The update amount of the bias of each neuron; Indicates the output layer of the previous iteration The update amount of the bias of each neuron; Represents the hidden layer The error term of Indicates that from the input layer neurons to the hidden layer The update amount of the weight of each neuron; Indicates that the last iteration from the input layer neurons to the hidden layer The update amount of the weight of each neuron; Represents the hidden layer The update amount of the bias of each neuron; Indicates the last iteration of the hidden layer The update amount of the bias of each neuron;

[0048] S135. Update weights and biases according to the calculated gradients:

[0049] ;

[0050] ;

[0051] ;

[0052] ;

[0053] S136. Use the training set to iterate steps S132 to S135 until a stopping condition is met.

[0054] Preferably, the performance evaluation indicators described in step S15 include R value, MAE, MAPE and RMSE.

[0055] The invention discloses a vibration load sliding slope section tunnel impact simulation test device, which is used for executing a vibration load sliding slope section tunnel impact simulation test method.

[0056] Therefore, the present invention adopts the above-mentioned vibration load sliding slope section tunnel impact simulation test method and device, which has the following beneficial effects:

[0057] 1. Accurate prediction model: By constructing a multi-hazard damage prediction model, multiple factors (such as earthquake parameters, tunnel landslide parameters, etc.) can be comprehensively considered to improve the accuracy of prediction;

[0058] Utilization of historical data: Utilizing historical earthquake parameters and tunnel landslide parameters, the model can better reflect the actual situation and improve the reliability of prediction;

[0059] 2. Experimental verification: Through actual testing on a vibration table, the tunnel landslide damage test data is collected, and then the predicted data is compared with the actual test data. The boundary conditions are adjusted based on the comparison results to obtain scientific boundary conditions to provide support for subsequent experiments;

[0060] 3. Dynamic adjustment: If the deviation between the predicted data and the actual test data exceeds the threshold, the model can be automatically optimized by adjusting the boundary conditions and retesting.

[0061] Iterative improvement: Through repeated adjustments and tests, the prediction accuracy of the model is gradually improved to ensure the reliability of the final results;

[0062] 4. Data-driven: By calculating the degree of deviation between predicted data and actual test data, decisions can be made based on data to avoid the uncertainty of subjective judgment;

[0063] 5. Similarity theory: Through similarity theory, the experimental results in the laboratory can be extended to actual engineering, thus improving the application value of the research results;

[0064] Parametric design: The test model is constructed based on the tunnel landslide test parameters, and parametric design can be performed according to different engineering requirements, with high flexibility;

[0065] 6. Visual output: Data display: By collecting tunnel landslide damage data and outputting it for display, the experimental results can be displayed intuitively, which is convenient for analysis and reporting;

[0066] Correspondence table: Output the correspondence table between the current boundary conditions and tunnel landslide parameters to provide reference for subsequent research;

[0067] 7. Risk assessment: Based on the experimental results, the design of the tunnel can be optimized to improve the safety and stability of the project.

[0068] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 The present invention is a flow chart of a tunnel impact simulation test method in a landslide section under a vibration load. DETAILED DESCRIPTION

[0070] In the description of the present invention, it should be noted that the terms "upper", "lower", "inside", "outside", etc. indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, or the positions or positional relationships in which the invented product is usually placed when in use. They are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. In the description of the present invention, it should also be noted that, unless otherwise clearly specified and limited, the terms "setting", "installation", and "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be a connection between the two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0071] The embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0072] like Figure 1 As shown, a vibration load tunnel impact simulation test method in a landslide section includes the following steps:

[0073] S1. Construct a multi-hazard damage prediction model, and input historical earthquake parameters and tunnel landslide parameters into the multi-hazard damage prediction model to obtain tunnel landslide damage prediction data;

[0074] Step S1 specifically includes the following steps:

[0075] S11, using the collected historical earthquake parameters and tunnel landslide parameters to form a data set, and dividing the data set into a training set and a test set according to the ratio;

[0076] Earthquake parameters include earthquake acceleration ,earthquake magnitude , epicenter distance and earthquake duration ; Tunnel landslide parameters include landslide hazard index , Geological Strength Index and damage index ; Among them, the earthquake acceleration ,earthquake magnitude , epicenter distance , Earthquake duration , Landslide Hazard Index and geological intensity index are the input vectors of the multi-hazard damage prediction model , damage index is the output vector of the multi-hazard damage prediction model ;

[0077] S12. A multi-hazard damage prediction model is constructed using a feedforward neural network algorithm, wherein the multi-hazard damage prediction model is expressed as follows:

[0078] ;

[0079] In the formula, represents the distance coefficient; represents the terrain factor; An index indicating the magnitude of an earthquake; Represents earthquake acceleration and earthquake magnitude Index of Represents terrain factor Index of

[0080] The multi-hazard damage prediction model described in step S12 includes an input layer, three hidden layers and an output layer; and uses a logarithmic S-type transfer function as a nonlinear activation function of the hidden layer neurons, and uses a linear transfer function as a transfer function of the output layer.

[0081] In step S12, Take 0.003, Take 3.9, Take 5.3.

[0082] S13, inputting the training set into the multi-hazard damage prediction model, training the multi-hazard damage prediction model using the Levenberg-Marquardt back propagation algorithm and the gradient descent weight and bias learning function with momentum, repeatedly assigning the weight of each neuron through the training set to establish a correlation between the input and output layers of the multi-hazard damage prediction model;

[0083] Step S13 specifically includes the following steps:

[0084] S131, initializing a multi-hazard damage prediction model;

[0085] S132, forward propagation: input the training set into the multi-hazard damage prediction model and calculate the output of each layer;

[0086] The output expression from the input layer to the hidden layer is as follows:

[0087] ;

[0088] ;

[0089] In the formula, Indicates The net input to the hidden layer neurons; Represents the first neurons to the hidden layer The connection weights of neurons; Represents the input layer The input features of each neuron; represents the hidden layer The bias of each neuron; represents the hidden layer The output of a neuron; represents the number of neurons in the input layer; represents the activation function;

[0090] The output expression from the hidden layer to the output layer is as follows:

[0091] ;

[0092] ;

[0093] In the formula, Indicates The net input to the output layer neurons; represents the number of neurons in the hidden layer; Indicates that from the hidden layer The first neuron to the output layer The connection weights of neurons; The output layer The bias of each neuron; The output layer The output of each neuron is considered as the network prediction output;

[0094] S133, calculate the target output of each sample in the training set and the network prediction output The difference between :

[0095] ;

[0096] S134. Back propagation: computing differences Gradients with respect to weights;

[0097] Among them, the difference from the output layer to the hidden layer The formula for calculating the gradient of weight is as follows:

[0098] ;

[0099] ;

[0100] ;

[0101] Differences between hidden layer and input layer The formula for calculating the gradient of weight is as follows:

[0102] ;

[0103] ;

[0104] ;

[0105] In the formula, Represents the output layer The error term of each neuron; Represents the activation function exist The derivative at ; Indicates that from the hidden layer neurons to the output layer The update amount of the weight of each neuron; represents the learning rate; represents the momentum term coefficient; Indicates that the last iteration from the hidden layer neurons to the output layer The weight update amount of each neuron; Represents the output layer The update amount of the bias of each neuron; Indicates the output layer of the previous iteration The update amount of the bias of each neuron; Represents the hidden layer The error term of Indicates that from the input layer neurons to the hidden layer The update amount of the weight of each neuron; Indicates that the last iteration from the input layer neurons to the hidden layer The update amount of the weight of each neuron; Represents the hidden layer The update amount of the bias of each neuron; Indicates the last iteration of the hidden layer The update amount of the bias of each neuron;

[0106] S135. Update weights and biases according to the calculated gradients:

[0107] ;

[0108] ;

[0109] ;

[0110] ;

[0111] S136. Use the training set to iterate steps S132 to S135 until a stopping condition is met.

[0112] S14, inputting the test set into the multi-hazard damage prediction model to test the performance of the multi-hazard damage prediction model;

[0113] S15. Evaluate the performance of the multi-hazard damage prediction model using performance evaluation indicators.

[0114] The performance evaluation indicators described in step S15 include R value, MAE, MAPE and RMSE.

[0115] S2, based on the same tunnel landslide parameters as step S1, construct a tunnel landslide test model, then place the tunnel landslide test model on a shaking table, and then based on the initial boundary conditions, input the same earthquake parameters as step S1 into the shaking table to collect tunnel landslide damage test data;

[0116] S3, calculating the deviation between the tunnel landslide damage prediction data obtained in step S1 and the tunnel landslide damage test data collected in step S2, and judging whether the deviation is greater than a threshold value, if so, adjusting the boundary conditions and returning to step S2; if not, outputting the corresponding relationship between the current boundary conditions and the tunnel landslide parameters to obtain a corresponding relationship table;

[0117] S4. Obtain tunnel landslide test parameters according to similarity theory, and construct a tunnel landslide test model based on the tunnel landslide test parameters. At the same time, based on the tunnel landslide test parameters, search the corresponding relationship table described in step S3 to obtain boundary conditions. Based on the obtained boundary conditions, input earthquake parameters to the shaking table, collect tunnel landslide damage data and output them for display.

[0118] The invention discloses a vibration load sliding slope section tunnel impact simulation test device, which is used for executing a vibration load sliding slope section tunnel impact simulation test method.

[0119] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for simulating the impact of tunnels on a landslide section under vibration load, characterized in that: The following steps are involved: S1. Use feedforward neural network algorithm to build a multi-hazard damage prediction model, and input historical earthquake parameters and tunnel landslide parameters into the multi-hazard damage prediction model to obtain tunnel landslide damage prediction data; S11, using the collected historical earthquake parameters and tunnel landslide parameters to form a data set, and dividing the data set into a training set and a test set according to the ratio; Earthquake parameters include earthquake acceleration ,earthquake magnitude , epicenter distance and earthquake duration ; Tunnel landslide parameters include landslide hazard index , Geological Strength Index and damage index ; Among them, the earthquake acceleration ,earthquake magnitude , epicenter distance , Earthquake duration , Landslide Hazard Index and geological intensity index are the input vectors of the multi-hazard damage prediction model , damage index is the output vector of the multi-hazard damage prediction model ; S12. A multi-hazard damage prediction model is constructed using a feedforward neural network algorithm, wherein the multi-hazard damage prediction model is expressed as follows: ； In the formula, represents the distance coefficient; represents the terrain factor; An index indicating the magnitude of an earthquake; Represents earthquake acceleration and earthquake magnitude Index of Represents terrain factor Index of S13, inputting the training set into the multi-hazard damage prediction model, training the multi-hazard damage prediction model using the Levenberg-Marquardt back propagation algorithm and the gradient descent weight and bias learning function with momentum, repeatedly assigning the weight of each neuron through the training set to establish a correlation between the input and output layers of the multi-hazard damage prediction model; S2, based on the same tunnel landslide parameters as step S1, construct a tunnel landslide test model, then place the tunnel landslide test model on a shaking table, and then based on the initial boundary conditions, input the same earthquake parameters as step S1 into the shaking table to collect tunnel landslide damage test data; S3, calculating the deviation between the tunnel landslide damage prediction data obtained in step S1 and the tunnel landslide damage test data collected in step S2, and judging whether the deviation is greater than a threshold value, if so, adjusting the boundary conditions and returning to step S2; if not, outputting the corresponding relationship between the current boundary conditions and the tunnel landslide parameters to obtain a corresponding relationship table; S4. Obtain tunnel landslide test parameters according to similarity theory, and construct a tunnel landslide test model based on the tunnel landslide test parameters. At the same time, based on the tunnel landslide test parameters, search the corresponding relationship table described in step S3 to obtain boundary conditions. Based on the obtained boundary conditions, input earthquake parameters to the shaking table, collect tunnel landslide damage data and output them for display.

2. The method for simulating the impact of a tunnel on a landslide under vibration load according to claim 1 is characterized by: The multi-hazard damage prediction model described in step S12 includes an input layer, three hidden layers and an output layer; And use the logarithmic sigmoid transfer function as the nonlinear activation function of the hidden layer neurons, and use the linear transfer function as the transfer function of the output layer; After step S13, the following steps are also included: S14, inputting the test set into the multi-hazard damage prediction model to test the performance of the multi-hazard damage prediction model; S15. Evaluate the performance of the multi-hazard damage prediction model using performance evaluation indicators.

3. The method for simulating the impact of a tunnel on a landslide under vibration load according to claim 2 is characterized by: In step S12, Take 0.003, Take 3.9, Take 5.

3.

4. The method for simulating the impact of a tunnel on a landslide under vibration load according to claim 3 is characterized by: Step S13 specifically includes the following steps: S131, initializing a multi-hazard damage prediction model; S132, forward propagation: input the training set into the multi-hazard damage prediction model and calculate the output of each layer; The output expression from the input layer to the hidden layer is as follows: ； ； In the formula, Indicates The net input to the hidden layer neurons; Represents the first neurons to the hidden layer The connection weights of neurons; Represents the input layer The input features of each neuron; represents the hidden layer The bias of each neuron; represents the hidden layer The output of a neuron; represents the number of neurons in the input layer; represents the activation function; The output expression from the hidden layer to the output layer is as follows: ； ； In the formula, Indicates The net input to the output layer neurons; represents the number of neurons in the hidden layer; Indicates that from the hidden layer The neurons to the output layer The connection weights of neurons; The output layer The bias of each neuron; The output layer The output of each neuron is considered as the network prediction output; S133, calculate the target output of each sample in the training set and the network prediction output The difference between : ； S134. Back propagation: computing differences Gradients with respect to weights; Among them, the difference from the output layer to the hidden layer The formula for calculating the gradient of weight is as follows: ； ； ； Differences between hidden layer and input layer The formula for calculating the gradient of weight is as follows: ； ； ； In the formula, Represents the output layer The error term of each neuron; Represents the activation function exist The derivative at ; Indicates that from the hidden layer neurons to the output layer The update amount of the weight of each neuron; represents the learning rate; represents the momentum term coefficient; Indicates that the last iteration from the hidden layer neurons to the output layer The weight update amount of each neuron; Represents the output layer The update amount of the bias of each neuron; Indicates the output layer of the previous iteration The update amount of the bias of each neuron; Represents the hidden layer The error term of Indicates that from the input layer neurons to the hidden layer The update amount of the weight of each neuron; Indicates that the last iteration from the input layer neurons to the hidden layer The update amount of the weight of each neuron; Represents the hidden layer The update amount of the bias of each neuron; Indicates the last iteration of the hidden layer The update amount of the bias of each neuron; S135. Update weights and biases according to the calculated gradients: ； ； ； ； S136. Use the training set to iterate steps S132 to S135 until a stopping condition is met.

5. The method for simulating the impact of tunnels on a landslide section under vibration load according to claim 4 is characterized by: The performance evaluation indicators described in step S15 include R value, MAE, MAPE and RMSE.

Images