A reinforced retaining wall reinforcement failure determination method based on apparent displacement monitoring

Abstract

The application discloses a reinforced retaining wall reinforcement failure determination method based on apparent displacement monitoring, and comprises the following steps: S1, determining the apparent displacement of the monitoring point of the reinforced retaining wall on site, and performing on-site investigation; S2, screening, classifying, interpolating and integrating by using the Kriging method; S3, constructing a theoretical model of the reinforced retaining wall, and determining the theoretical distribution range of the elastic modulus of the reinforcement material; S4, performing numerical simulation analysis on the theoretical model of the reinforced retaining wall, and obtaining the corresponding analysis result of the target displacement; S5, establishing a neural network model, and learning the mapping relationship between the layered reinforcement material elastic modulus and the surface monitoring point displacement; S6, obtaining the elastic modulus of the layered reinforcement material through the neural network model inversion; and S7, analyzing the relative relationship between the elastic modulus of the reinforcement material and the elastic modulus of the soil, and predicting whether the pull-out failure occurs. The application can accurately capture the actual working state of the retaining wall, and realizes comprehensive evaluation and failure prediction on the stability of the reinforced retaining wall.

Description

Technical Field

[0001] This invention relates to the field of civil engineering operation and maintenance technology, and in particular to a method for determining the failure of reinforced retaining walls based on apparent displacement monitoring. Background Technology

[0002] Reinforced retaining walls, as an economical and efficient retaining structure, are widely used in infrastructure construction such as roads, railways, and water conservancy. Structural safety and stability evaluation during the service life of reinforced retaining walls is of great significance for the reinforcement and maintenance of embankment slopes and ensuring the safe operation of roads. The failure modes of reinforced retaining walls mainly include two categories: external stability failure and internal stability failure. External failure generally manifests as overall structural instability and overturning; internal failure mainly manifests as tensile failure, pull-out failure of the reinforcing materials, and connection failure at the joint between the reinforcing materials and the panel. In actual engineering, the failure of reinforced retaining walls often presents as a comprehensive failure, with multiple destructive effects occurring simultaneously and coupling with each other.

[0003] Currently, methods for analyzing and predicting the stability of reinforced retaining walls mainly fall into two categories: one is based on physical and mechanical models, and the other is based on monitoring data feedback analysis. Methods based on physical and mechanical models primarily include the limit equilibrium method and the finite element method. The limit equilibrium method analyzes the strength of the reinforcing material required for the overall structural stability failure, i.e., stress distribution. For flexible reinforced retaining walls, the most commonly used method is the anchored wedge method. Compared to the limit equilibrium method, the finite element method can calculate the displacement, stress, and strain levels at various points in the soil, and can simulate the construction process and consider complex boundary conditions, thus more accurately assessing the stability of reinforced soil retaining walls and providing a scientific basis for engineering design, construction, and operation. However, the parameters of the finite element method require complex experiments to determine, and the difficulty in quantitatively judging failure limits its applicability in engineering.

[0004] The monitoring data feedback analysis method, unlike the physical mechanics model solution method, primarily focuses on model learning to analyze and predict the structural stability of reinforced retaining walls. However, traditional neural network models have certain limitations. For example, BP neural networks use static modeling, which is less effective for highly nonlinear and time-dependent data. Summary of the Invention

[0005] To address the challenges of analyzing and predicting the stability of reinforced retaining walls using existing methods based on physical mechanics models and monitoring data feedback, this invention proposes a method for determining the failure of reinforced retaining walls based on apparent displacement monitoring.

[0006] This application discloses a method for determining the reinforcement failure of reinforced retaining walls based on apparent displacement monitoring, including the following steps:

[0007] S1. Determine the apparent displacement of the monitoring points of the reinforced retaining wall on site, and conduct on-site investigation of the location of the reinforced retaining wall to obtain the physical and mechanical parameters of the surrounding soil and the structural mechanical parameters of the retaining wall.

[0008] S2. Based on the physical and mechanical parameters of the surrounding soil and geological data obtained from the field survey, the Kriging method is used for screening, classification, interpolation and integration.

[0009] S3. Based on the arrangement and distribution of reinforcement in the retaining wall structure system, determine the theoretical distribution range of the elastic modulus of the reinforcement, generate an estimate of the elastic modulus of the reinforcement using random sampling in a uniform distribution, and complete the characteristic distribution based on its arrangement and distribution in the reinforced retaining wall.

[0010] S4. Based on the on-site geological characteristics of the reinforced retaining wall after interpolation, the physical parameters of the reinforcement material generated by random sampling are used as the variation characteristics to construct a numerical simulation model of the retaining wall and conduct numerical simulation analysis to obtain the corresponding analysis results of the target displacement.

[0011] S5. Establish a neural network model, analyze the results of several monitoring points in the actual construction site corresponding to the numerical simulation model of the retaining wall, learn the mapping relationship between the elastic modulus of the layered reinforcement and the displacement of the surface monitoring point, until the neural network model passes the test.

[0012] S6. Import the measured surface displacement data of the reinforced retaining wall into the tested neural network model, and obtain the elastic modulus of the layered reinforcement through model inversion.

[0013] S7. Feed back the elastic modulus of the reinforcement to the actual measurement points in the project, and analyze the relative relationship between the elastic modulus of the reinforcement and the elastic modulus of the soil in combination with the actual distribution of the reinforcement, so as to predict whether the geogrid will fail by pull-out.

[0014] Preferably, step S1 includes the following steps:

[0015] S11. Determine the monitoring points for the reinforced retaining wall on-site, install equipment including earth pressure cells and flexible displacement gauges, and mark the monitoring points.

[0016] S12. Conduct geological surveys on the slope soil within the area affecting the reinforced retaining wall, select multiple survey points, and obtain the physical and mechanical parameters of the surrounding soil and the structural mechanical parameters of the retaining wall, including the elastic modulus of the reinforcement, soil density, soil elastic modulus, Poisson's ratio of the soil, interfacial cohesion between the soil and the reinforcement, and interfacial friction angle between the soil and the reinforcement.

[0017] Preferably, step S2 includes the following steps:

[0018] S21. Integrate the physical and mechanical parameters of the surrounding soil, including soil density and elastic properties, and mark the coordinates of the corresponding survey points;

[0019] S22. According to the first law of geography, select points within the effective influence range of Kriging interpolation as points to be interpolated, and mark the coordinates of the points to be interpolated.

[0020] S23. Determine the mathematical form of Kriging interpolation, and let 0 be the point to be interpolated. For the surrounding survey points, the estimated value of a certain regional material parameter at point 0 is:

[0021]

[0022] in, Indicates material parameters at The estimated value at that location, Indicates material parameters at Observations at that location These are the weighting coefficients. For Lagrange daily numbers, For the material to be located in position The higher-order trend function at the location, The correlation coefficient is the higher-order trend function. This represents the number of higher-order trend functions;

[0023] To minimize the difference between the expected value and the true value of the material parameters at the interpolation point, i.e.:

[0024]

[0025]

[0026] in, The values ​​are the actual material parameters at the insertion point. These are estimated values ​​of the material parameters at the insertion point;

[0027] S24, Calculation The estimated values ​​of the distance between the group of material parameters and the strain difference are used to construct the Kriging coefficient matrix based on a certain class of material parameters at a single interpolation point.

[0028]

[0029] The extended form of the Kriging interpolation matrix for the material to be interpolated is as follows:

[0030]

[0031] in, Point to be inserted At a known point The variation between material parameters, The covariance matrix of the variable itself. The covariance matrix is ​​the matrix between different variables;

[0032] S25. After calculating the material parameter distance and the estimated strain difference, calculate the estimated value of the variation function using the following formula:

[0033]

[0034] in, The spacing is The total number of all point pairs, To and Point deviation The measured value of the variable;

[0035] S26, will The distances and strain differences between data points are divided into several groups, each containing... For each set of distance values, calculate the average distance for that set. and mean variation We selected the Gaussian model as the theoretical model for the variogram, fitted the variogram and regressed the parameters of the Gaussian model to obtain the fitted variogram model.

[0036] S27. Substitute the distance between the survey point and the point to be interpolated into the fitted variation function to calculate the variation value of the point to be interpolated. And substitute all the variation values ​​into the formula In the process, the weight coefficients of the insertion points are calculated. Then according to the formula Obtain the true values ​​of the material parameters at the point to be interpolated;

[0037] S28. Repeat S25~S27 until the material parameter values ​​of all selected interpolation points are determined sequentially.

[0038] S29. Repeat S25~S28 until the material parameters for all selected categories are determined sequentially.

[0039] Preferably, step S3 includes the following steps:

[0040] S31. Determine the arrangement and distribution of reinforcement materials in the retaining wall structure system, including quantity, horizontal and vertical spacing, and anchorage length;

[0041] S32. Ensure that the overall sample of the elastic modulus of a given reinforcing material follows a uniform distribution. The sample parameters of the elastic modulus of the rib after integration and interpolation are as follows: ;

[0042] S33. Using the Gauss-Cauchy mixed density estimation method, the uniform distribution of the elastic modulus of the reinforcing bar is solved. The kernel density estimate is expressed as:

[0043]

[0044] in, Point The probability density estimate of the elastic modulus at that point. This represents the number of samples for the elastic modulus of the rib. For the first The elastic modulus of a single rib sample For the bandwidth of the Gaussian kernel, For the bandwidth of the Cauchy core, The weighting coefficients for the Gaussian kernel density estimation. The weighting coefficients for the Cauchy kernel density estimation;

[0045] bandwidth The optimal value is calculated according to the Silverman rule, that is:

[0046]

[0047] in, This represents the standard deviation of the elastic modulus of the ribs.

[0048] S34. Analyze the results of kernel density estimation. By plotting and observing the kernel density estimation map, identify the regions where the density curve is significantly higher than zero, determine the support interval of the data, i.e., the effective range of the data distribution, and take the lower limit of this interval as the uniform distribution parameter of the ribs. The value of is taken as the upper limit of the interval as the uniform distribution parameter of the ribs. The value;

[0049] S35. Based on the solved parameters, determine the uniform distribution of the elastic modulus of the reinforcing bar. Random sampling is performed on this uniform distribution using the sinusoidal-linear composite congruence method, as shown in the following formula:

[0050]

[0051] in, The first random number sequence to be determined for the next step One value, Indicates the currently determined number A random number, As a multiplier, For increments, For modulus, The weighting coefficients represent the influence of the sine function on the random number sequence. The frequency parameter of the sine function;

[0052] Determine the upper and lower boundaries of the distribution, and map the normalized random numbers obtained by sampling to the distribution interval in order to determine the elastic modulus of the reinforcement.

[0053] S36. Arrange the elastic modulus of the randomly selected reinforcement materials according to their arrangement and distribution pattern in the reinforced retaining wall to complete the characteristic distribution.

[0054] Preferably, step S4 includes the following steps:

[0055] S41. Based on the physical and mechanical parameters of the surrounding soil and the structural mechanical parameters of the retaining wall obtained from the field survey in S1, establish the three-dimensional model of the retaining wall in sequence, determine the analysis steps for the simulation model calculation of the retaining wall, define the contact conditions of the surrounding soil, the retaining wall and the reinforcement, create load and constraint conditions and perform mesh generation, complete the preliminary establishment of the numerical simulation model of the retaining wall, submit the results and generate the inp file.

[0056] S42. Construct an Abaqus random field model based on the interpolated soil material parameters and the randomly extracted reinforcement elastic modulus parameters, assign the soil material parameters to the mesh elements, and assign the reinforcement elastic modulus to the reinforcement component material.

[0057] S43. Write a programming script to generate inp files in batches based on the results of the random field and import them into the Abaqus software. Set up a batch calculation task to run all the generated inp files and obtain the target displacement analysis results.

[0058] S44. Compile the elastic modulus parameter values ​​of the reinforcement and the corresponding target displacement results in all inp files, and determine whether the reinforced retaining wall has failed based on the displacement.

[0059] Preferably, step S5 includes the following steps:

[0060] S51. Use the KNN weighted nearest neighbor algorithm to remove and clean the outliers caused by calculation errors in the elastic modulus parameter values ​​and the corresponding target displacement values ​​of the retaining wall in S4.

[0061] S52. Calculate the local weighted anomaly factor :

[0062]

[0063] in, For the standardized elastic modulus corresponding to the apparent displacement data points The average distance of the nearest neighbors For the standardized elastic modulus corresponding to the apparent displacement data points Average distance between data points For comprehensive weighting factors;

[0064]

[0065] in, This is the weighting factor bias coefficient. For distance-based weighting factors, This is a weighting factor for the correlation between elastic modulus and apparent displacement;

[0066] S53. Define an outlier discrimination index by combining locally weighted outlier factors and standardized data. :

[0067]

[0068] in, For weight parameters, This is a weighted average of the standardized apparent displacement distances. This represents the absolute deviation of the median of the standardized apparent displacement distance.

[0069] If the outlier threshold is three times the absolute deviation of the median, then if... Then it is believed If a point is an outlier, find the nearest non-outlier value to replace it;

[0070] S54. The cleaned elastic modulus corresponding to the apparent displacement data are used to form a dataset. The dataset is divided into a training set and a test set. A deep support vector machine regression model is established, and the radial basis function is used as the kernel function.

[0071] Selecting the optimal support vector regression parameters using Bayesian optimization methods: regularization parameters , and error tolerance The value ranges of the three types of parameters are determined sequentially.

[0072] The negative mean square error of the apparent displacement observation results is used as the optimization objective:

[0073]

[0074] in, To test the true value of the lumped apparent displacement, To test the predicted values ​​of the lumped apparent displacements, The number of samples in the test set;

[0075] S55. Select Gaussian process EI as the surrogate model and use the acquisition function to select the next optimization point:

[0076]

[0077] Repeat the optimization process until the optimal parameters with the lowest mean square error are found.

[0078] S56. Construct a deep support vector regression model, using a multi-layer support vector regression structure to capture the data pattern of apparent displacement corresponding to the elastic modulus. The model output is expressed as follows:

[0079]

[0080] in, This represents the number of layers in a deep support vector regression model. For radial basis functions and kernel functions, For the weight of each layer, For bias terms;

[0081] S57. Train the deep support vector regression model using the training set. Use the optimal parameters determined by Bayesian optimization to train the deep support vector regression model. Introduce a regularization term to prevent overfitting. The objective function is shown in the following equation:

[0082]

[0083] in, For the weight vector, and As slack variables, To control the hyperparameters of regularization strength, For regularization terms;

[0084] The objective function optimization is constrained by the following conditions:

[0085]

[0086] in, For kernel function mapping, The input is a randomly generated set of elastic moduli for reinforcing bars. The set of output monitoring displacement observation results is defined, and the upper and lower limits of the prediction results are specified. The objective function is obtained by fitting a deep support vector regression model.

[0087] S58. Apply the model trained in S57 to the test set for prediction, targeting... The accuracy of the depth support vector regression model was assessed by evaluating the displacement observation of reinforced soil using a set of test samples and by measuring the mean square error and mean absolute error.

[0088] S59. Analyze the results of mean squared error and mean absolute error to evaluate the predictive performance of the deep support vector regression model. If either the mean squared error or the mean absolute error is higher than 10, then modify the parameter range and Bayesian optimization surrogate model using the methods in S54-S55 to further correct the parameters of the deep support vector regression model until the deep support vector regression model has good predictive accuracy.

[0089] Preferably, step S6 includes the following steps:

[0090] S61. Obtain displacement data of the soil measured on site using a flexible displacement gauge, and check the integrity and continuity of the data;

[0091] S62. For abnormal displacement data, the KNN weighted nearest neighbor algorithm of S51-S53 is used for data cleaning. The boundary of outliers is calculated according to the degree of variability and outliers are identified and replaced.

[0092] S63. Construct a feature matrix based on the cleaned displacement data, use a deep support vector regression model to predict the input displacement data, and output the corresponding predicted value of the elastic modulus of the reinforcing bar.

[0093] Preferably, step S7 includes the following steps:

[0094] If the elastic modulus of the reinforcement and the elastic modulus of the soil satisfy any of the following formulas, then the prediction results indicate that the geogrid will experience pull-out failure:

[0095]

[0096]

[0097]

[0098] in, For the effective reinforcement system number, The elastic modulus of the rib. The elastic modulus of the soil. This refers to the amount of reinforcement in the transverse direction of the retaining wall. This refers to the quantity of reinforcement in the longitudinal direction of the retaining wall. This refers to the reinforcement objects located at any position in the horizontal row of the retaining wall. This refers to the reinforcement object located at any position in the longitudinal direction of the retaining wall.

[0099] Effective reinforcement system number Satisfy the following formula:

[0100]

[0101] in, This is the actual length of the rib. This is the effective length of the rib.

[0102] The beneficial effects of this invention are:

[0103] (1) Based on on-site monitoring and geological survey, this invention accurately captures the actual working state of the retaining wall by collecting apparent displacement data from on-site monitoring points and combining it with first-hand soil physical and mechanical parameters obtained from geological survey, and provides effective data support for prediction and analysis.

[0104] (2) This invention uses Kriging interpolation to spatially interpolate geological data, effectively extending local soil properties to the overall area affecting the stability of retaining walls, and improving the continuity and integrity of the spatial distribution of parameters.

[0105] (3) The present invention simulates the actual distribution of the elastic modulus of the reinforcement material by random sampling method, and constructs a numerical simulation model of the retaining wall, so that the failure of the reinforcement is more in line with the laws of physics.

[0106] (4) The present invention introduces a neural network model based on SVR, which realizes accurate inversion of the elastic modulus of the reinforcement, thereby predicting the stability of the retaining wall.

[0107] (5) The present invention realizes a comprehensive assessment of the stability of reinforced retaining walls and predicts their failure, providing a new technical means for the stability assessment and failure early warning of reinforced retaining walls. Attached Figure Description

[0108] Figure 1 This is a schematic diagram of the process for determining the reinforcement failure of a reinforced retaining wall based on apparent displacement monitoring, according to an embodiment of the present invention.

[0109] Figure 2 This is a diagram showing the layout of instruments at geogrid monitoring points according to an embodiment of the present invention.

[0110] Figure 3 This is a schematic diagram of the actual reinforcement layout of the reinforced retaining wall according to an embodiment of the present invention;

[0111] Figure 4 This is a flowchart illustrating the implementation of predicting the elastic modulus using a deep support vector regression model in an embodiment of the present invention. Detailed Implementation

[0112] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided with reference to the accompanying drawings and embodiments.

[0113] This invention discloses a method for determining the failure of reinforced retaining walls based on apparent displacement monitoring, the process of which is as follows: Figure 1 As shown, it includes the following steps:

[0114] S1. Determine the apparent displacement of the monitoring points of the reinforced retaining wall on site, and conduct on-site investigation of the location of the reinforced retaining wall. Based on physical exploration, drilling and other means, obtain the physical and mechanical parameters of the surrounding soil and the structural mechanical parameters of the retaining wall in a targeted manner.

[0115] S11. Determine the monitoring points for the reinforced retaining wall on-site, deploy equipment including earth pressure cells and flexible displacement gauges, and mark the monitoring points. In this embodiment, the monitoring points are as follows: Figure 2 As shown, a total of [number] settings are available. Layers, number of layers Based on the actual engineering requirements, several flexible displacement gauges (>3) will be installed on each floor, staggered between floors. Earth pressure cells will be installed directly above the monitoring points on each floor, near the back of the reinforced retaining wall, totaling [number missing]. Each monitoring point is labeled with its geometric location, apparent displacement, and specific earth pressure value.

[0116] S12. Conduct a geological survey of the slope soil within the area affecting the reinforced retaining wall, selecting multiple survey points. In this embodiment, the survey points are randomly selected directly above the soil on the back side of the reinforced retaining wall, with a planar distance of not less than 20cm between the points. The survey mainly obtains the physical and mechanical parameters of the surrounding soil and the structural mechanical parameters of the retaining wall, including but not limited to the elastic modulus of the reinforcement, soil density, soil elastic modulus, Poisson's ratio, interfacial cohesion between the soil and reinforcement, and interfacial friction angle between the soil and reinforcement.

[0117] S2. Based on the physical and mechanical parameters of the surrounding soil and geological data obtained from the on-site investigation, the Kriging method is used for screening, classification, interpolation, and integration. Geological data interpolation employs the Kriging method, extending the material properties of the local soil on the embankment slope to the overall portion affecting the stability of the retaining wall.

[0118] S21. Integrate the physical and mechanical parameters of the surrounding soil, including soil density and elastic properties, and mark the coordinates of the corresponding survey points.

[0119] S22. According to the first law of geography, select points within the effective influence range of Kriging interpolation as the points to be interpolated, and mark the coordinates of the points to be interpolated.

[0120] S23. Determine the mathematical form of Kriging interpolation, and let 0 be the point to be interpolated. For the surrounding survey points, the estimated value of a certain regional material parameter at point 0 is:

[0121]

[0122] in, Indicates material parameters at The estimated value at that location, Indicates material parameters at Observations at that location These are the weighting coefficients. This represents the calculated regionalized material parameters. For Lagrange daily numbers, For the material to be located in position The higher-order trend function at the location, The correlation coefficient is the higher-order trend function. This represents the number of higher-order trend functions.

[0123] To ensure unbiased estimation, the difference between the expected value and variance of the estimated value and the true value of the material parameter at the interpolation point should be minimized, i.e., the following condition must be met:

[0124]

[0125]

[0126] in, The values ​​are the actual material parameters at the insertion point. These are estimated values ​​for the material parameters at the insertion point.

[0127] S24, Calculation The estimated values ​​of the distance between the group of material parameters and the strain difference are used to construct the Kriging coefficient matrix based on a certain class of material parameters at a single interpolation point.

[0128]

[0129] Based on this matrix, considering the cross-correlation among multiple material variables, the extended form of the material interpolation point kriging interpolation matrix is ​​constructed as follows:

[0130]

[0131] in, Point to be inserted At a known point The variation between material parameters, The covariance matrix of the variable itself. This is the covariance matrix between different variables.

[0132] S25. After calculating the material parameter distance and the estimated strain difference, calculate the estimated value of the variation function using the following formula:

[0133]

[0134] in, Distance The variation value, The spacing is The total number of all point pairs, To and Point deviation The measured value of the variable.

[0135] S26, will The distances and strain differences between data points are divided into several groups, each containing... For each set of distance values, calculate the average distance for that set. and mean variation The Gaussian model is chosen as the theoretical model for the variogram, and the expression for the variogram is:

[0136]

[0137] in, This represents the error value caused by data variability over short distances. This represents the theoretical maximum value of the variation function. The range of the variation function. To adjust the curve smoothness parameters.

[0138] The variation function is fitted and the parameters of the Gaussian model are regressed to obtain the fitted variation function model.

[0139] S27. Substitute the distance between the survey point and the point to be interpolated into the fitted variation function to calculate the variation value of the point to be interpolated. And substitute all the variation values ​​into the formula In the process, the weight coefficients of the insertion points are calculated. Then according to the formula The true values ​​of the material parameters of the point to be interpolated are obtained.

[0140] S28. Repeat S25~S27 until the material parameter values ​​of all selected interpolation points are determined sequentially.

[0141] S29. Repeat S25~S28 until the material parameters for all selected categories are determined sequentially.

[0142] S3. Based on the arrangement and distribution of reinforcement materials in the retaining wall structure system, determine the theoretical distribution range of the elastic modulus of the reinforcement materials, generate an estimate of the elastic modulus of the reinforcement materials using random sampling in a uniform distribution, and complete the characteristic distribution based on their arrangement and distribution in the reinforced retaining wall.

[0143] S31, such as Figure 3 As shown, the arrangement and distribution of reinforcing bars in the retaining wall structural system are determined, including the quantity, longitudinal and transverse spacing, and anchorage length. The number of reinforcing bars in the figure is... The anchorage length and the horizontal and vertical spacing are determined by the actual site conditions. The reinforcement bars are arranged at equal intervals in the horizontal direction and at unequal intervals in the vertical direction.

[0144] S32. Ensure that the overall sample of the elastic modulus of a given reinforcing material follows a uniform distribution. The sample parameters of the elastic modulus of the rib after integration and interpolation are as follows: .

[0145] S33. Using the Gauss-Cauchy mixed density estimation method, the uniform distribution of the elastic modulus of the reinforcing bar is solved. The kernel density estimate is expressed as:

[0146]

[0147] in, Point The probability density estimate of the elastic modulus at that point. This represents the number of samples for the elastic modulus of the rib. For the first The elastic modulus of a single rib sample For the bandwidth of the Gaussian kernel, For the bandwidth of the Cauchy core, The weighting coefficients for the Gaussian kernel density estimation. These are the weighting coefficients for the Cauchy kernel density estimation.

[0148] bandwidth The optimal value is calculated according to the Silverman rule, that is:

[0149]

[0150] in, This represents the standard deviation of the elastic modulus of the ribs.

[0151] S34. Analyze the results of kernel density estimation. By plotting and observing the kernel density estimation map, identify the regions where the density curve is significantly higher than zero, determine the support interval of the data, i.e., the effective range of the data distribution, and take the lower limit of this interval as the uniform distribution parameter of the ribs. The value of is taken as the upper limit of the interval as the uniform distribution parameter of the ribs. The value of .

[0152] S35. Based on the solved parameters, determine the uniform distribution of the elastic modulus of the reinforcing bar. Random sampling is performed on this uniform distribution using the sinusoidal-linear composite congruence method, as shown in the following formula:

[0153]

[0154] in, The first random number sequence to be determined for the next step One value, Indicates the currently determined number A random number, As a multiplier, For increments, and Using the Park-Miller constant, the given model is a high-parameter system. In this embodiment, the modulus is used. Values or , The weighting coefficients represent the influence of the sine function on the random number sequence. is the frequency parameter of the sine function.

[0155] Determine the upper and lower boundaries of the distribution, and map the normalized random numbers obtained by sampling to the distribution interval to determine the elastic modulus of the reinforcing bar.

[0156] S36. Arrange the elastic modulus of the randomly selected reinforcement materials according to their arrangement and distribution pattern in the reinforced retaining wall to complete the characteristic distribution.

[0157] S4. Based on the on-site geological characteristics of the interpolated reinforced retaining wall, the physical parameters of the randomly sampled reinforcement are used as the variation characteristics to construct a numerical simulation model of the retaining wall and perform numerical simulation analysis to obtain the corresponding analysis results of the target displacement.

[0158] S41. Based on the physical and mechanical parameters of the surrounding soil and the structural mechanical parameters of the retaining wall obtained from the field survey in S1, establish the three-dimensional model of the retaining wall in sequence, determine the analysis steps for the simulation model calculation of the retaining wall, define the contact conditions of the surrounding soil, the retaining wall and the reinforcement, create load and constraint conditions and perform mesh generation, complete the preliminary establishment of the numerical simulation model of the retaining wall, submit the results and generate the inp file.

[0159] S42. Construct an Abaqus random field model based on the interpolated soil material parameters and the randomly extracted reinforcement elastic modulus parameters. Assign soil material parameters to mesh elements and reinforcement elastic modulus to reinforcement component materials.

[0160] S43. Write a programming script to generate .inp files in batches based on the results of the random field and import them into the Abaqus software. Set up a batch calculation task to run all the generated .inp files and obtain the target displacement analysis results.

[0161] S44. Compile the elastic modulus parameter values ​​of the reinforcement and the corresponding target displacement results in all inp files. Refer to the reinforced soil retaining wall specifications to determine whether the reinforced retaining wall has failed based on the displacement.

[0162] S5, such as Figure 4As shown, a neural network model is established, and the results of several monitoring points in the actual construction site corresponding to the numerical simulation model of the retaining wall are analyzed. The mapping relationship between the elastic modulus of the layered reinforcement and the displacement of the surface monitoring points is learned until the neural network model passes the test.

[0163] S51. The outliers in the elastic modulus parameter values ​​and corresponding retaining wall target displacement values ​​from S4, caused by calculation errors, are removed and cleaned using the KNN weighted nearest neighbor algorithm. First, the elastic modulus data... and apparent displacement data Standardization is performed, as shown in equation (13), to obtain the standardized elastic modulus data. and apparent displacement data Assume that the two sets of data correspond one-to-one to form two-dimensional coordinate points ( ), ( ), ……( ), traverse any two points in two-dimensional space ( )and( ), perform weighted KNN distance calculation:

[0164]

[0165]

[0166] in, Let be the coordinates of any point in two-dimensional space. Let be the coordinates of any other point in two-dimensional space. Let be the Euclidean distance between any two points in two-dimensional space. This is a distance-based weighting factor.

[0167]

[0168] in, The standardized elastic modulus, For standardized apparent displacement data, This represents the mean of the elastic modulus and apparent displacement. For the standard deviation of elastic modulus and apparent displacement.

[0169] The cosine similarity is used to calculate the similarity measure between each point, thus obtaining a weighting factor for the correlation between the elastic modulus and the apparent displacement:

[0170]

[0171] in, This represents the average value of the standardized elastic modulus sample data. The average value of the standardized apparent displacement data.

[0172] S52. Calculate the local weighted anomaly factor :

[0173]

[0174] in, For the standardized elastic modulus corresponding to the apparent displacement data points The average distance of the nearest neighbors For the standardized elastic modulus corresponding to the apparent displacement data points Average distance between data points This is a comprehensive weighting factor.

[0175]

[0176] in, This is the weighting factor bias coefficient, with a value ranging from 0 to 1.

[0177] S53. Define an outlier discrimination index by combining locally weighted outlier factors and standardized data. :

[0178]

[0179] in, This is a weighting parameter, with a value ranging from 0 to 1. This is a weighted average of the standardized apparent displacement distances. This represents the absolute deviation of the median of the standardized apparent displacement distance.

[0180] If the outlier threshold is three times the absolute deviation of the median, then if... Then it is believed If a point is an outlier, find the nearest non-outlier value to replace it.

[0181] S54. The cleaned elastic modulus corresponding to the apparent displacement data are used to form a dataset. The dataset is divided into a training set and a test set. A deep support vector machine regression model is established, and the radial basis function (RBF) is used as the kernel function.

[0182] Selecting the optimal Support Vector Regression (SVR) parameters using Bayesian optimization methods: Regularization parameters , and error tolerance The value ranges of the three types of parameters are determined sequentially: .

[0183] The negative mean square error of the apparent displacement observation results is used as the optimization objective:

[0184]

[0185] in, To test the true value of the lumped apparent displacement, To test the predicted values ​​of the lumped apparent displacements, This represents the number of samples in the test set.

[0186] S55. Select Gaussian process EI as the surrogate model and use the acquisition function to select the next optimization point:

[0187]

[0188] Repeat the optimization process until the optimal parameters with the lowest mean square error are found.

[0189] S56. Construct a deep support vector regression model, using a multi-layer support vector regression structure to capture the data pattern of apparent displacement corresponding to the elastic modulus. The model output is expressed as follows:

[0190]

[0191] in, This represents the number of layers in a deep support vector regression model. For radial basis functions and kernel functions, For the weight of each layer, This is a bias term.

[0192] S57. Train the deep support vector regression model using the training set. Use the optimal parameters determined by Bayesian optimization to train the deep support vector regression model. Introduce a regularization term to prevent the model from overfitting. The objective function is shown in the following equation:

[0193]

[0194] in, For the weight vector, and As slack variables, To control the hyperparameters of regularization strength, This is a regularization term.

[0195] The objective function optimization is constrained by the following conditions:

[0196]

[0197] in, For kernel function mapping, The input is a randomly generated set of elastic moduli for reinforcing bars. This is the set of output monitoring displacement observation results. Upper and lower bound constraints are defined for the prediction results, i.e., the maximum and minimum values ​​of historical monitoring displacement data are taken as the initial upper and lower bounds. Finally, a deep support vector regression model is fitted to obtain the objective function.

[0198] S58. Apply the model trained in S57 to the test set for prediction, targeting... The accuracy of the depth support vector regression model for reinforced soil displacement observation was evaluated using a set of test samples and mean squared error (MSE) and mean absolute error (MAE).

[0199]

[0200]

[0201] in, For the first The actual displacement observation results of each sample For the first Predicted displacement observation results for each sample.

[0202] S59. Analyze the results of mean squared error and mean absolute error to evaluate the predictive performance of the deep support vector regression model. If either MSE or MAE is higher than 10, modify the parameter range and Bayesian optimization surrogate model using the methods in S54-S55 to further correct the parameters of the deep support vector regression model until the error between the MSE and MAE values ​​after the last fitting and the MSE and MAE values ​​after the previous fitting of the deep support vector regression model does not exceed 10%, indicating that the deep support vector regression model has good predictive accuracy.

[0203] S6, such as Figure 4 As shown, the surface displacement data of the reinforced retaining wall measured on site is imported into a tested neural network model, and the elastic modulus of the layered reinforcement is obtained through model inversion.

[0204] S61. Obtain displacement data of the soil measured on site using a flexible displacement gauge, and check the integrity and continuity of the data.

[0205] S62. For abnormal displacement data, the KNN weighted nearest neighbor algorithm (S51-S53) is used for data cleaning. The boundary of outliers is calculated based on the degree of variability, outliers are identified, and outliers are replaced.

[0206] S63. Construct a feature matrix based on the cleaned displacement data, use a deep support vector regression model to predict the input displacement data, and output the corresponding predicted value of the elastic modulus of the reinforcing bar.

[0207] S7. Feed back the elastic modulus of the reinforcement to the actual measurement points in the project, and analyze the relative relationship between the elastic modulus of the reinforcement and the elastic modulus of the soil in combination with the actual distribution of the reinforcement, so as to predict whether the geogrid will fail by pull-out.

[0208] If the elastic modulus of the reinforcement and the elastic modulus of the soil satisfy any of the following formulas, then the prediction results indicate that the geogrid will experience pull-out failure:

[0209]

[0210]

[0211]

[0212] in, For the effective reinforcement system number, The elastic modulus of the rib. The elastic modulus of the soil. This refers to the amount of reinforcement in the transverse direction of the retaining wall. This refers to the quantity of reinforcement in the longitudinal direction of the retaining wall. This refers to the reinforcement objects located at any position in the horizontal row of the retaining wall. This refers to the reinforcement object located at any position in the longitudinal direction of the retaining wall.

[0213] Effective reinforcement system number Satisfy the following formula:

[0214]

[0215] in, This is the actual length of the rib. This is the effective length of the rib.

[0216] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

Claims

1. A method for determining the failure of reinforced retaining walls based on apparent displacement monitoring, characterized in that, Includes the following steps: S1. Determine the apparent displacement of the monitoring points of the reinforced retaining wall on site, and conduct on-site investigation of the location of the reinforced retaining wall to obtain the physical and mechanical parameters of the surrounding soil and the structural mechanical parameters of the retaining wall. S2. Based on the physical and mechanical parameters of the surrounding soil and geological data obtained from the field survey, the Kriging method is used for screening, classification, interpolation and integration. S3. Based on the arrangement and distribution of reinforcement in the retaining wall structure system, determine the theoretical distribution range of the elastic modulus of the reinforcement, generate an estimate of the elastic modulus of the reinforcement using random sampling in a uniform distribution, and complete the characteristic distribution based on its arrangement and distribution in the reinforced retaining wall. S4. Based on the on-site geological characteristics of the reinforced retaining wall after interpolation, the physical parameters of the reinforcement material generated by random sampling are used as the variation characteristics to construct a numerical simulation model of the retaining wall and conduct numerical simulation analysis to obtain the corresponding analysis results of the target displacement. S5. Establish a neural network model, analyze the results of several monitoring points in the actual construction site corresponding to the numerical simulation model of the retaining wall, learn the mapping relationship between the elastic modulus of the layered reinforcement and the displacement of the surface monitoring point, until the neural network model passes the test. S6. Import the measured surface displacement data of the reinforced retaining wall into the tested neural network model, and obtain the elastic modulus of the layered reinforcement through model inversion. S7. Feedback the elastic modulus of the reinforcement to the actual measurement points in the project, and analyze the relative relationship between the elastic modulus of the reinforcement and the elastic modulus of the soil in combination with the actual distribution of the reinforcement, so as to predict whether the geogrid will experience pull-out failure. This includes the following steps: If the elastic modulus of the reinforcement and the elastic modulus of the soil satisfy any of the following formulas, then the prediction results indicate that the geogrid will experience pull-out failure: in, For the effective reinforcement system number, The elastic modulus of the rib. The elastic modulus of the soil. This refers to the amount of reinforcement in the transverse direction of the retaining wall. This refers to the quantity of reinforcement in the longitudinal direction of the retaining wall. This refers to the reinforcement objects located at any position in the horizontal row of the retaining wall. This refers to the reinforcement object located at any column position in the longitudinal direction of the retaining wall; Effective reinforcement system number Satisfy the following formula: in, This is the actual length of the rib. This is the effective length of the rib.

2. The method for determining the reinforcement failure of a reinforced retaining wall based on apparent displacement monitoring according to claim 1, characterized in that, S1 includes the following steps: S11. Determine the monitoring points for the reinforced retaining wall on-site, install equipment including earth pressure cells and flexible displacement gauges, and mark the monitoring points. S12. Conduct geological surveys on the slope soil within the area affecting the reinforced retaining wall, select multiple survey points, and obtain the physical and mechanical parameters of the surrounding soil and the structural mechanical parameters of the retaining wall, including the elastic modulus of the reinforcement, soil density, soil elastic modulus, Poisson's ratio of the soil, interfacial cohesion between the soil and the reinforcement, and interfacial friction angle between the soil and the reinforcement.

3. The method for determining the reinforcement failure of a reinforced retaining wall based on apparent displacement monitoring according to claim 2, characterized in that, S2 includes the following steps: S21. Integrate the physical and mechanical parameters of the surrounding soil, including soil density and elastic properties, and mark the coordinates of the corresponding survey points; S22. According to the first law of geography, select points within the effective influence range of Kriging interpolation as points to be interpolated, and mark the coordinates of the points to be interpolated. S23. Determine the mathematical form of Kriging interpolation, and let 0 be the point to be interpolated. For the surrounding survey points, the estimated value of a certain regional material parameter at point 0 is: in, Indicates material parameters at The estimated value at that location, Indicates material parameters at Observations at that location These are the weighting coefficients. For Lagrange daily numbers, For the material to be located in position The higher-order trend function at the location, The correlation coefficient is the higher-order trend function. This represents the number of higher-order trend functions; To minimize the difference between the expected value and the true value of the material parameters at the interpolation point, i.e.: in, The values ​​are the actual material parameters at the insertion point. These are estimated values ​​of the material parameters at the insertion point; S24, Calculation The estimated values ​​of the distance between the group of material parameters and the strain difference are used to construct the Kriging coefficient matrix based on a certain class of material parameters at a single interpolation point. The extended form of the Kriging interpolation matrix for the material to be interpolated is as follows: in, Point to be inserted At a known point The variation between material parameters, The covariance matrix of the variable itself. The covariance matrix is ​​the matrix between different variables; S25. After calculating the material parameter distance and the estimated strain difference, calculate the estimated value of the variation function using the following formula: in, The spacing is The total number of all point pairs, To and Point deviation The measured value of the variable; S26, will The distances and strain differences between data points are divided into several groups, each containing... For each set of distance values, calculate the average distance for that set. and mean variation We selected the Gaussian model as the theoretical model for the variogram, fitted the variogram and regressed the parameters of the Gaussian model to obtain the fitted variogram model. S27. Substitute the distance between the survey point and the point to be interpolated into the fitted variation function to calculate the variation value of the point to be interpolated. And substitute all the variation values ​​into the formula In the process, the weight coefficients of the insertion points are calculated. Then according to the formula Obtain the true values ​​of the material parameters at the point to be interpolated; S28. Repeat S25~S27 until the material parameter values ​​of all selected interpolation points are determined sequentially. S29. Repeat S25~S28 until the material parameters for all selected categories are determined sequentially.

4. The method for determining the reinforcement failure of a reinforced retaining wall based on apparent displacement monitoring according to claim 3, characterized in that, S3 includes the following steps: S31. Determine the arrangement and distribution of reinforcement materials in the retaining wall structure system, including quantity, horizontal and vertical spacing and anchorage length; S32. Ensure that the overall sample of the elastic modulus of a given reinforcing material follows a uniform distribution. The sample parameters of the elastic modulus of the rib after integration and interpolation are as follows: ; S33. Using the Gauss-Cauchy mixed density estimation method, the uniform distribution of the elastic modulus of the reinforcing bar is solved. The kernel density estimate is expressed as: in, Point The probability density estimate of the elastic modulus at that point. This represents the number of samples for the elastic modulus of the rib. For the first The elastic modulus of a single rib sample For the bandwidth of the Gaussian kernel, For the bandwidth of the Cauchy core, The weighting coefficients for the Gaussian kernel density estimation. The weighting coefficients for the Cauchy kernel density estimation; bandwidth The optimal value is calculated according to the Silverman rule, that is: in, This represents the standard deviation of the elastic modulus of the ribs. S34. Analyze the results of the kernel density estimation, determine the support interval of the data, and take the upper limit of this interval as the uniform distribution parameter of the ribs. The value; S35. Based on the solved parameters, determine the uniform distribution of the elastic modulus of the reinforcing bar. Random sampling is performed on this uniform distribution using the sinusoidal-linear composite congruence method, as shown in the following formula: in, The first random number sequence to be determined for the next step One value, Indicates the currently determined number A random number, As a multiplier, For increments, For modulus, The weighting coefficients represent the influence of the sine function on the random number sequence. The frequency parameter of the sine function; Determine the upper and lower boundaries of the distribution, and map the normalized random numbers obtained by sampling to the distribution interval in order to determine the elastic modulus of the reinforcement. S36. Arrange the elastic modulus of the randomly selected reinforcement materials according to their arrangement and distribution pattern in the reinforced retaining wall to complete the characteristic distribution.

5. The method for determining the reinforcement failure of a reinforced retaining wall based on apparent displacement monitoring according to claim 4, characterized in that, S4 includes the following steps: S41. Based on the physical and mechanical parameters of the surrounding soil and the structural mechanical parameters of the retaining wall obtained from the field survey in S1, establish the three-dimensional model of the retaining wall in sequence, determine the analysis steps for the simulation model calculation of the retaining wall, define the contact conditions of the surrounding soil, the retaining wall and the reinforcement, create load and constraint conditions and perform mesh generation, complete the preliminary establishment of the numerical simulation model of the retaining wall, submit the results and generate the inp file. S42. Construct an Abaqus random field model based on the interpolated soil material parameters and the randomly extracted reinforcement elastic modulus parameters, assign the soil material parameters to the mesh elements, and assign the reinforcement elastic modulus to the reinforcement component material. S43. Write a programming script to generate inp files in batches based on the results of the random field and import them into the Abaqus software. Set up a batch calculation task to run all the generated inp files and obtain the target displacement analysis results. S44. Compile the elastic modulus parameter values ​​of the reinforcement and the corresponding target displacement results in all inp files, and determine whether the reinforced retaining wall has failed based on the displacement.

6. The method for determining the reinforcement failure of a reinforced retaining wall based on apparent displacement monitoring according to claim 5, characterized in that, S5 includes the following steps: S51. Use the KNN weighted nearest neighbor algorithm to remove and clean the outliers caused by calculation errors in the elastic modulus parameter values ​​and the corresponding target displacement values ​​of the retaining wall in S4. S52. Calculate the local weighted anomaly factor : in, For the standardized elastic modulus corresponding to the apparent displacement data points The average distance of the nearest neighbors For the standardized elastic modulus corresponding to the apparent displacement data points Average distance between data points For comprehensive weighting factors; in, This is the weighting factor bias coefficient. For distance-based weighting factors, This is a weighting factor for the correlation between elastic modulus and apparent displacement; S53. Define an outlier discrimination index by combining locally weighted outlier factors and standardized data. : in, For weight parameters, This is a weighted average of the standardized apparent displacement distances. This represents the absolute deviation of the median of the standardized apparent displacement distance. If the outlier threshold is three times the absolute deviation of the median, then if... Then it is believed If a point is an outlier, find the nearest non-outlier value to replace it; S54. The cleaned elastic modulus corresponding to the apparent displacement data are used to form a dataset. The dataset is divided into a training set and a test set. A deep support vector machine regression model is established, and the radial basis function is used as the kernel function. Selecting the optimal support vector regression parameters using Bayesian optimization methods: regularization parameters , and error tolerance The value ranges of the three types of parameters are determined sequentially. The negative mean square error of the apparent displacement observation results is used as the optimization objective: in, To test the true value of the lumped apparent displacement, To test the predicted values ​​of the lumped apparent displacements, The number of samples in the test set; S55. Select Gaussian process EI as the surrogate model and use the acquisition function to select the next optimization point: Repeat the optimization process until the optimal parameters with the lowest mean square error are found. S56. Construct a deep support vector regression model, using a multi-layer support vector regression structure to capture the data pattern of apparent displacement corresponding to the elastic modulus. The model output is expressed as follows: in, This represents the number of layers in a deep support vector regression model. For radial basis functions and kernel functions, For the weight of each layer, For bias terms; S57. Train the deep support vector regression model using the training set. Use the optimal parameters determined by Bayesian optimization to train the deep support vector regression model. Introduce a regularization term to prevent overfitting. The objective function is shown in the following equation: in, For the weight vector, and As slack variables, To control the hyperparameters of regularization strength, For regularization terms; The objective function optimization is constrained by the following conditions: in, For kernel function mapping, The input is a randomly generated set of elastic moduli for reinforcing bars. The set of output monitoring displacement observation results is defined, and the upper and lower limits of the prediction results are specified. The objective function is obtained by fitting a deep support vector regression model. S58. Apply the model trained in S57 to the test set for prediction, targeting... The accuracy of the depth support vector regression model was assessed by evaluating the displacement observation of reinforced soil using a set of test samples and by measuring the mean square error and mean absolute error. S59. Analyze the results of mean squared error and mean absolute error to evaluate the predictive performance of the deep support vector regression model. Modify the parameter range of mean squared error and mean absolute error using the methods in S54-S55 and optimize the surrogate model with Bayesian optimization to further correct the parameters of the deep support vector regression model until the deep support vector regression model has good predictive accuracy.

7. The method for determining the reinforcement failure of a reinforced retaining wall based on apparent displacement monitoring according to claim 6, characterized in that, S6 includes the following steps: S61. Obtain displacement data of the soil measured on site using a flexible displacement gauge, and check the integrity and continuity of the data; S62. For abnormal displacement data, the KNN weighted nearest neighbor algorithm of S51-S53 is used for data cleaning. The boundary of outliers is calculated according to the degree of variability and outliers are identified and replaced. S63. Construct a feature matrix based on the cleaned displacement data, use a deep support vector regression model to predict the input displacement data, and output the corresponding predicted value of the elastic modulus of the reinforcing bar.

Images