Slope pile-anchor supporting structure optimization design method based on deep learning

Abstract

The invention discloses a deep learning-based side slope pile-anchor support structure optimization design method, which comprises the following steps of: obtaining different side slope pile-anchor support structure parameters and dynamic response data under the action of an earthquake working condition; preprocessing the data, dividing a training set and a test set, and reconstructing time sequence data; constructing a prediction model based on a bidirectional long-short-term memory network, and combining with the improved virus colony search to optimize the hyper-parameters of the prediction model; and inputting data in the training set to train the model to obtain an optimal prediction model, predicting a dynamic response result of the pile-anchor supporting structure by using the prediction model, and optimizing and adjusting a design scheme of the pile-anchor supporting structure according to the result. According to the method, the design time is greatly shortened, the design cost is greatly reduced, the action results of different seismic dynamic loads on the pile-anchor supporting structure can be known more meticulously, the reliability of the design results is improved, and the method has good social and economic benefits and engineering application prospects.

Description

Technical Field

[0001] The present invention relates to the technical field of slope support structure design, and in particular to a slope pile-anchor support structure optimization design method based on deep learning. Background Art

[0002] Compared with traditional single anti-sliding piles or anchoring structures, pile-anchor support structures have superior seismic performance and are widely used. However, due to their structural complexity, their dynamic response under seismic loads is more nonlinear than that of traditional single support structures, which increases the design difficulty.

[0003] In the current seismic design of slope engineering, the dynamic response analysis using time-history analysis is an important method for seismic design of slope engineering. However, due to the complexity of the problem itself, the mathematical and physical methods used in the traditional time-history analysis method are difficult to accurately describe the strong nonlinear relationship between seismic loads and dynamic responses. At the same time, due to the complex and changeable characteristics of seismic loads, it is time-consuming and expensive to carry out large-scale centrifugal vibration table model tests under multiple working conditions for a single project, and the design funds are difficult to meet. Using the previous centrifugal vibration table model test data and the monitoring data of the existing slope support structure, the dynamic response law of the pile-anchor support structure under different seismic loads is obtained, which becomes the key to optimizing its structural design, reducing design costs, and shortening design time.

[0004] In recent years, artificial intelligence methods have made significant progress. By utilizing complex algorithms and large data sets, artificial intelligence methods have shown great advantages in modeling complex nonlinear relationships and adaptively handling dynamic interactions of uncertain systems. In particular, deep learning has shown significant advantages in processing complex time series analysis data. Introducing deep learning technology into the optimization design of slope pile-anchor support structures can make full use of field monitoring data and historical engineering data, establish a nonlinear mapping relationship between slope response and support parameters, solve the limitations of traditional methods, and improve the scientificity and accuracy of the design. It can achieve efficient evaluation and optimization of different geological conditions and support structure design schemes, and finally determine the best support structure design scheme, providing a scientific, accurate and efficient design tool for slope support engineering. Summary of the invention

[0005] The purpose of the present invention is to provide a deep learning-based optimization design method for slope pile-anchor support structures. This method predicts the dynamic response results such as pile end displacement and anchor tension of the pile-anchor support structure under different seismic loads by learning the centrifugal shaking table model test data of previous pile-anchor support structures and the monitoring data of slope support structures.

[0006] To achieve the above functions, the present invention designs a slope pile-anchor support structure optimization design method based on deep learning. For the slope soil and the pile-anchor support structure, the following steps S1 to S5 are performed to complete the optimization of the slope pile-anchor support structure:

[0007] Step S1: Through centrifugal shaking table tests and on-site monitoring of slope pile-anchor support structures, dynamic response data of different slope pile-anchor support structures under earthquake motion, as well as corresponding slope soil strength parameters and geometric parameters, and pile-anchor support structure design parameters are obtained; a sample database of slopes and pile-anchor support structures is constructed;

[0008] Step S2: preprocessing the dynamic response data, including using discrete wavelet transform method to reduce noise and normalize the dynamic response data; dividing the sample database of slope and pile anchor support structure into 80% training set and 20% test set; reconstructing the dynamic response data by step reconstruction method;

[0009] Step S3: construct and train a deep learning prediction model based on an optimization algorithm to optimize hyperparameters;

[0010] Step S4: using earthquake motion data, slope soil strength parameters and geometric parameters, and pile-anchor support structure design parameters as input data of the deep learning prediction model, and using the preprocessed dynamic response data as output data of the deep learning prediction model;

[0011] By dividing the set into 80% training sets and 20% test sets and then using 5-fold cross-validation, the evaluation index of each validation is calculated to determine the optimal deep learning prediction model;

[0012] Step S5: Based on the optimal deep learning prediction model, the design parameters of the pile-anchor support structure, as well as the strength parameters and geometric parameters of the slope soil, are input, and according to the seismic protection design objectives, the seismic motion data are input to predict the dynamic response results of the pile-anchor support structure under the action of seismic motion; according to the dynamic response results output by the deep learning model, the design scheme of the pile-anchor support structure is optimized and adjusted.

[0013] Beneficial effects: Compared with the prior art, the advantages of the present invention include:

[0014] The present invention designs a slope pile-anchor support structure optimization design method based on deep learning. The method can provide an evaluation for the parameter design of the pile-anchor support structure and optimize and adjust the design parameters according to the dynamic response results. Compared with the traditional centrifuge model test for design result verification, the design time and design cost are greatly shortened, and the effects of different seismic loads on the pile-anchor support structure can be understood in more detail. The reliability of the design results is improved, with good social and economic benefits and engineering application prospects. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a flow chart of a slope pile-anchor support structure optimization design method based on deep learning provided according to an embodiment of the present invention;

[0016] Figure 2 is a schematic diagram of a step-by-step reconstruction method provided according to an embodiment of the present invention;

[0017] Figure 3 is a structural diagram of a deep learning prediction model provided according to an embodiment of the present invention;

[0018] Figure 4 is a schematic diagram of deep learning prediction model training convergence provided according to an embodiment of the present invention;

[0019] Figure 5 This is a graph of 5-fold cross-validation results of a deep learning prediction model provided according to an embodiment of the present invention;

[0020] Figure 6 is a model diagram of a pile-anchor support model test provided according to an embodiment of the present invention;

[0021] Figure 7 This is a graph of pile body bending moment prediction results of the deep learning prediction model provided according to an embodiment of the present invention. DETAILED DESCRIPTION

[0022] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0023] The embodiment of the present invention provides a slope pile anchor support structure optimization design method based on deep learning, for slope soil and pile anchor support structure, referring to Figure 1 , perform the following steps S1 to S5 to complete the optimization of the slope pile-anchor support structure:

[0024] Step S1: Through centrifugal shaking table tests and on-site monitoring of slope pile-anchor support structures, dynamic response data of different slope pile-anchor support structures under earthquake motion, as well as corresponding slope soil strength parameters and geometric parameters, and pile-anchor support structure design parameters are obtained; a sample database of slopes and pile-anchor support structures is constructed;

[0025] The slope soil strength parameters include soil elastic modulus E, Poisson's ratio μ, gravity γ, cohesion c and internal friction angle φ; the slope geometric parameters include slope height and slope angle; the slope pile anchor support structure dynamic response data include anchor angle and diameter, pile length, width and thickness, pile anchor joint coordinates, pile body elastic modulus E p , elastic modulus of anchor rod E a , time t, anchor tension T and pile top displacement D.

[0026] Step S2: preprocessing the dynamic response data, including using discrete wavelet transform method to reduce noise and normalize the dynamic response data; dividing the sample database of slope and pile anchor support structure into 80% training set and 20% test set; reconstructing the dynamic response data by step reconstruction method;

[0027] The discrete wavelet transform method reduces the high-frequency components of the dynamic response data, thereby highlighting the low-frequency components in the data, which helps to achieve signal smoothing and improve the deep learning prediction model to identify and analyze the long-term changes or periodic trends of the dynamic response data. The outputs of the high-frequency and low-frequency components of the discrete wavelet transform method are:

[0028]

[0029] x=x hp +x lp

[0030] Where X is the dynamic response input data of the original pile-anchor support structure, x hp and x lp are the outputs of high-frequency and low-frequency components respectively, and are the filtering parameters for adjusting the proportion of high-frequency and low-frequency components, and x is the output data after noise reduction.

[0031] The normalization is normalization within the range of (-1,1), as shown below:

[0032]

[0033] Among them, x n is the normalized data; x min and x max are the minimum and maximum values ​​of the entire feature variable in the data set; y min and max are the minimum and maximum values ​​in each data group, respectively.

[0034] Reference Figure 2 The step reconstruction method selects 10% of the data set as the step length and 1 data as the step distance, so as to reconstruct the dynamic response data of the slope pile-anchor support structure in the form of 2×1000 time series into a reconstructed data set of 901×2×100; that is, the time series data with a length of 1000 is expanded into 901 groups of sequences, and the length of each group of sequences is 100.

[0035] Step S3: construct and train a deep learning prediction model based on an optimization algorithm to optimize hyperparameters;

[0036] In step S3, a deep learning prediction model is constructed based on a bidirectional long short-term memory network. Figure 3 , the training model convergence process is as follows Figure 4 As shown; the bidirectional long short-term memory network parameter values ​​are shown in Table 1:

[0037] Table 1 Bidirectional long short-term memory network parameter values

[0038]

[0039] In one embodiment, the optimization algorithm based on the deep learning prediction model adopts an improved virus colony search algorithm (IVCS), and improves the immune response stage of the virus colony search algorithm (VCS), thereby avoiding the limitations of the virus colony search algorithm and enhancing its solution ability; the specific improvement method is to introduce a random process based on the mutation principle, and the specific process is to define a mutation rate mr ranging from 0 to 1 for a solution vector X with dimension n in the virus colony search algorithm; for each element x in the solution vector X i , generate a random number r in the range [0, 1]; if r < mr, then for x i Perform a mutation operation; the mutation operation is to i A random perturbation is added, which is generated by a normal distribution with mean 0 and a set standard deviation ms (mutation strength), and mr and ms are defined as 0.1 and 0.01 respectively.

[0040] The optimization algorithm in step S3 can be flexibly selected. Taking the dung beetle optimization algorithm as an example, the parameters of the optimization algorithm are set as shown in Table 2 below:

[0041] Table 2 Parameter settings of dung beetle optimization algorithm.

[0042]

[0043] In order to avoid unnecessary calculations, a stopping criterion is defined, that is, the root mean square error (RMSE) fluctuation of the objective function result within 10 consecutive iterations is less than 0.01. The value of the stopping criterion should be reasonably adjusted for different calculation scenarios. The calculation of RMSE is as follows:

[0044]

[0045] In the formula, n represents the amount of data; y i and are the predicted value and true value of the dynamic response of the i-th pile-anchor supporting structure.

[0046] In the process of training the deep learning model based on the optimization algorithm, the hyperparameters of the deep learning model are adjusted according to the loss function. The hyperparameters include the number of neurons in the hidden layer (n h), learning rate (LR), learning rate reduction factor and regularization coefficient (L2).

[0047] Step S4: using earthquake motion data, slope soil strength parameters and geometric parameters, and pile-anchor support structure design parameters as input data of the deep learning prediction model, and using the preprocessed dynamic response data as output data of the deep learning prediction model;

[0048] By dividing the set into 80% training sets and 20% test sets and then using 5-fold cross-validation, the evaluation index of each validation is calculated to determine the optimal deep learning prediction model;

[0049] The five-fold cross validation is used to evaluate the impact of different data partitions on model training, reflect the stability of the model through the fluctuation of multiple evaluation indicators, and determine the optimal deep learning prediction model; the evaluation indicators include regression coefficient (R 2 ), root mean square error (RMSE), absolute error (MAE), mean square error (MSE). Refer to the 5-fold cross-validation results of the deep learning prediction model Figure 5 .

[0050] Step S5: Based on the optimal deep learning prediction model, the design parameters of the pile-anchor support structure, as well as the strength parameters and geometric parameters of the slope soil, are input, and according to the seismic protection design objectives, the seismic motion data are input to predict the dynamic response results of the pile-anchor support structure under the action of seismic motion; according to the dynamic response results output by the deep learning model, the design scheme of the pile-anchor support structure is optimized and adjusted.

[0051] The embodiment of the present invention also provides a slope pile anchor support structure optimization design system based on deep learning, including:

[0052] one or more processors;

[0053] A memory storing operable instructions, wherein when the instructions are executed by the one or more processors, the one or more processors are caused to perform operations, and the following steps S1 to S5 are performed for the slope soil and the pile-anchor support structure to complete the optimization of the slope pile-anchor support structure:

[0054] Step S1: Through centrifugal shaking table tests and on-site monitoring of slope pile-anchor support structures, dynamic response data of different slope pile-anchor support structures under earthquake motion, as well as corresponding slope soil strength parameters and geometric parameters, and pile-anchor support structure design parameters are obtained; a sample database of slopes and pile-anchor support structures is constructed;

[0055] Step S2: preprocessing the dynamic response data, including using discrete wavelet transform method to reduce noise and normalize the dynamic response data; dividing the sample database of slope and pile anchor support structure into 80% training set and 20% test set; reconstructing the dynamic response data by step reconstruction method;

[0056] Step S3: construct and train a deep learning prediction model based on an optimization algorithm to optimize hyperparameters;

[0057] Step S4: using earthquake motion data, slope soil strength parameters and geometric parameters, and pile-anchor support structure design parameters as input data of the deep learning prediction model, and using the preprocessed dynamic response data as output data of the deep learning prediction model;

[0058] By dividing the set into 80% training sets and 20% test sets and then using 5-fold cross-validation, the evaluation index of each validation is calculated to determine the optimal deep learning prediction model;

[0059] Step S5: Pile-anchor support model test model diagram Figure 6 As shown in the figure; based on the optimal deep learning prediction model, the design parameters of the pile-anchor support structure, as well as the strength parameters and geometric parameters of the slope soil, are input, and according to the seismic protection design objectives, the seismic motion data is input to predict the dynamic response results of the pile-anchor support structure under the action of seismic motion; according to the dynamic response results output by the deep learning model, the design scheme of the pile-anchor support structure is optimized and adjusted. Taking the pile bending moment as an example, the seismic motion and dynamic response prediction results are input as shown in the figure. Figure 7 shown.

[0060] An embodiment of the present invention also provides a computer-readable medium storing software, wherein the computer-readable medium includes instructions that can be executed by one or more computers, and when the instructions are executed by the one or more computers, the operations of the method for optimizing the design of slope pile-anchor support structures based on deep learning are performed.

[0061] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge scope of ordinary technicians in this field without departing from the purpose of the present invention.

Claims

1. A deep learning-based optimization design method for slope pile-anchor support structure, characterized in that: For the slope soil and the pile-anchor support structure, the following steps S1 to S5 are performed to complete the optimization of the slope pile-anchor support structure: Step S1: Through centrifugal shaking table tests and on-site monitoring of slope pile-anchor support structures, dynamic response data of different slope pile-anchor support structures under earthquake motion, as well as corresponding slope soil strength parameters and geometric parameters, and pile-anchor support structure design parameters are obtained; a sample database of slopes and pile-anchor support structures is constructed; Step S2: preprocessing the dynamic response data, including using discrete wavelet transform method to reduce noise and normalize the dynamic response data; dividing the sample database of slope and pile anchor support structure into 80% training set and 20% test set; reconstructing the dynamic response data by step reconstruction method; Step S3: construct and train a deep learning prediction model based on an optimization algorithm to optimize hyperparameters; Step S4: using earthquake motion data, slope soil strength parameters and geometric parameters, and pile-anchor support structure design parameters as input data of the deep learning prediction model, and using the preprocessed dynamic response data as output data of the deep learning prediction model; By dividing the set into 80% training sets and 20% test sets and then using 5-fold cross-validation, the evaluation index of each validation is calculated to determine the optimal deep learning prediction model; Step S5: Based on the optimal deep learning prediction model, the design parameters of the pile-anchor support structure, as well as the strength parameters and geometric parameters of the slope soil, are input, and according to the seismic protection design objectives, the seismic motion data are input to predict the dynamic response results of the pile-anchor support structure under the action of the seismic motion; According to the dynamic response results output by the deep learning model, the design scheme of the pile-anchor support structure is optimized and adjusted.

2. The method for optimizing the slope pile-anchor support structure based on deep learning according to claim 1, characterized in that: The slope soil strength parameters include soil elastic modulus E, Poisson's ratio μ, gravity γ, cohesion c and internal friction angle φ; the slope geometric parameters include slope height and slope angle; the slope pile anchor support structure dynamic response data include anchor angle and diameter, pile length, width and thickness, pile anchor joint coordinates, pile body elastic modulus E p , elastic modulus of anchor rod E a , time t, anchor tension T and pile top displacement D.

3. The method for optimizing the slope pile-anchor support structure based on deep learning according to claim 1, characterized in that: The discrete wavelet transform method described in step S2 has the following outputs for high-frequency and low-frequency components: x=x hp +x lp Where X is the dynamic response input data of the original pile-anchor support structure, x hp and x lp are the outputs of high-frequency and low-frequency components respectively, and are the filtering parameters for adjusting the proportion of high-frequency and low-frequency components, and x is the output data after noise reduction.

4. The method for optimizing the slope pile-anchor support structure based on deep learning according to claim 1, characterized in that: The normalization described in step S2 is normalization within the range of (-1, 1), as shown in the following formula: Among them, x n is the normalized data; x min and x max are the minimum and maximum values ​​of the entire feature variable in the data set; y min and max are the minimum and maximum values ​​in each data group, respectively.

5. The method for optimizing the slope pile-anchor support structure based on deep learning according to claim 1, characterized in that: The step reconstruction method described in step S2 selects 10% of the data set as the step length and 1 data as the step distance, thereby reconstructing the dynamic response data of the slope pile-anchor support structure in the form of 2×1000 time series into a 901×2×100 reconstructed data set.

6. The method for optimizing the slope pile-anchor support structure based on deep learning according to claim 1, characterized in that: In step S3, a deep learning prediction model is constructed based on a bidirectional long short-term memory network. The optimization algorithm based on the deep learning prediction model adopts an improved virus colony search algorithm, and the immune response stage of the virus colony search algorithm is improved. The specific improvement method is to introduce A random process based on the mutation principle. The specific process is to define a mutation rate mr ranging from 0 to 1 for the solution vector X with dimension n in the virus colony search algorithm; for each element x in the solution vector X i , generate a random number r in the range [0, 1]; if r < mr, then for x i Perform a mutation operation; the mutation operation is to i A random perturbation is added, which is generated by a normal distribution with mean 0 and a set standard deviation ms, and mr and ms are defined as 0.1 and 0.01 respectively.

7. The method for optimizing the slope pile-anchor support structure based on deep learning according to claim 1, characterized in that: In the process of training the deep learning model based on the optimization algorithm in step S3, the hyperparameters of the deep learning model are adjusted according to the loss function, and the hyperparameters include the number of hidden layer neurons, the learning rate, the learning rate reduction factor and the regularization coefficient.

8. The method for optimizing the design of slope pile-anchor support structure based on deep learning according to claim 1, characterized in that: The evaluation indicators described in step S4 include regression coefficient, root mean square error, absolute error, and mean square error.

9. A slope pile anchor support structure optimization design system based on deep learning, characterized in that: include: one or more processors; A memory storing operable instructions, wherein when the instructions are executed by the one or more processors, the one or more processors are caused to perform operations, and the following steps S1 to S5 are performed for the slope soil and the pile-anchor support structure to complete the optimization of the slope pile-anchor support structure: Step S1: Through centrifugal shaking table tests and on-site monitoring of slope pile-anchor support structures, dynamic response data of different slope pile-anchor support structures under earthquake motion, as well as corresponding slope soil strength parameters and geometric parameters, and pile-anchor support structure design parameters are obtained; a sample database of slopes and pile-anchor support structures is constructed; Step S2: preprocessing the dynamic response data, including using discrete wavelet transform method to reduce noise and normalize the dynamic response data; dividing the sample database of slope and pile anchor support structure into 80% training set and 20% test set; reconstructing the dynamic response data by step reconstruction method; Step S3: construct and train a deep learning prediction model based on an optimization algorithm to optimize hyperparameters; Step S4: using earthquake motion data, slope soil strength parameters and geometric parameters, and pile-anchor support structure design parameters as input data of the deep learning prediction model, and using the preprocessed dynamic response data as output data of the deep learning prediction model; By dividing the set into 80% training sets and 20% test sets and then using 5-fold cross-validation, the evaluation index of each validation is calculated to determine the optimal deep learning prediction model; Step S5: Based on the optimal deep learning prediction model, the design parameters of the pile-anchor support structure, as well as the strength parameters and geometric parameters of the slope soil, are input, and according to the seismic protection design objectives, the seismic motion data are input to predict the dynamic response results of the pile-anchor support structure under the action of the seismic motion; According to the dynamic response results output by the deep learning model, the design scheme of the pile-anchor support structure is optimized and adjusted.

10. A computer-readable medium storing software, characterized in that: The readable medium includes instructions that can be executed by one or more computers, and when the instructions are executed by the one or more computers, the operations of the deep learning-based slope pile anchor support structure optimization design method as described in any one of claims 1-8 are performed.

Images