High fill roadbed settlement prediction method

Abstract

The invention provides a high fill roadbed settlement prediction method, and belongs to the technical field of engineering construction.The method comprises the steps that firstly, a large test box interface test system is established, and a 4 * 4 sensor monitoring network is arranged on the interface of filler and a foundation soil layer; the influence of different parameter combinations on interface mechanical properties is researched through a parameter change experiment system, and a parameter database is established; obtaining actual measurement data of typical engineering and determining a calibration coefficient; establishing an interface element model, and obtaining an interface friction angle, cohesion and rigidity parameters; calculating the deformation of each layer by adopting a layered accumulation method; introducing a time effect correction factor to consider consolidation characteristics; the slip influence is evaluated through an interface slip test, and the prediction deviation caused by neglecting the interface effect in a traditional method is corrected; and finally, constructing a deep learning prediction model, and verifying prediction precision and reliability through actually measured data.

Description

Technical Field

[0001] The present invention belongs to the technical field of engineering construction, and more particularly, relates to a method for predicting settlement of high embankment subgrades. Background Art

[0002] High embankment subgrades are common engineering structure forms in the construction of linear projects such as highways and railways. Their settlement characteristics directly affect the safety and stability of the projects. Traditional methods for predicting settlement of high embankment subgrades mainly include empirical formula method, layered analysis method, finite element numerical simulation method, etc. The empirical formula method establishes empirical relationships based on a large amount of engineering practice data, which is easy to operate but has poor universality; the layered analysis method is based on Boussinesq stress theory, regards the foundation as an elastic half-space, and calculates the sum of the compression amounts of each layer; the finite element numerical simulation method solves the deformation field by establishing a calculation model and considering the nonlinear characteristics of materials.

[0003] However, these traditional methods all show obvious limitations in practical applications. First, the existing methods overly simplify the interaction mechanism between the filling and the foundation soil layer interface, and ignore the possible slip effect and stress concentration phenomenon at the interface; second, the traditional methods fail to fully consider the combined influence of filling parameters (such as density, water content, gradation, etc.) and foundation soil layer parameters; in addition, the existing prediction models usually simplify the time effect, and it is difficult to accurately reflect the settlement time history evolution law under the dual effects of compression and consolidation.

[0004] Especially under high embankment conditions, due to the large thickness of the filling, complex loads, and uneven stress distribution, the mechanical behavior at the interface between the filling and the foundation soil layer is more complex. The existing prediction methods are difficult to accurately characterize this complex process, resulting in a large deviation between the prediction results and the actual monitoring values, and unable to provide a reliable technical basis for engineering design and construction, and even may lead to potential engineering safety hazards. That is to say, there is a technical problem of insufficient accuracy in predicting settlement of high embankment subgrades in the prior art. Summary of the Invention

[0005] In view of this, the present invention provides a method for predicting settlement of high embankment subgrades, which can solve the technical problem of insufficient accuracy in predicting settlement of high embankment subgrades existing in the prior art.

[0006] The present invention is implemented as follows: The present invention provides a method for predicting the settlement of a high-fill subgrade, including: establishing an interface test system for the high-fill subgrade, installing displacement sensors and stress sensors at the interface between the filler and the foundation soil layer to form a monitoring network; performing an experiment on changing test parameters, and establishing a parameter database for the filler and the foundation soil layer; obtaining measured data of typical projects to obtain a calibration coefficient; establishing an interface element model for the filler and the foundation soil layer to obtain interface mechanical parameters; using the layered cumulative calculation method to calculate the total settlement; introducing a time effect correction factor to calculate the settlement time history prediction curve; conducting an evaluation test on the influence of interface slip to correct the prediction deviation value caused by ignoring the interface effect in the traditional settlement calculation method; constructing a settlement intelligent prediction system to generate a settlement prediction model for the high-fill subgrade; implementing subgrade settlement monitoring and verification, and outputting a settlement prediction report for the high-fill subgrade.

[0007] Among them, the establishment of the interface test system for the high-fill subgrade includes constructing a large indoor test box with dimensions of 1000×500×500 mm, filling test fillers and simulating the foundation soil layer, installing 16 high-precision displacement sensors and 16 stress sensors at the contact interface between the filler and the foundation soil layer to form a 4×4 monitoring network, and measuring the stress distribution value and deformation amount at the interface under the self-weight of the filler.

[0008] Among them, the implementation of the experiment on changing test parameters includes changing the density index of the test filler, the water content index of the test filler, the gradation composition index of the test filler, and the bearing capacity index of the foundation soil layer, setting the above indexes to 3 different values respectively, and measuring the friction coefficient parameter, the cohesion strength parameter, and the shear parameter at the interface between the filler and the foundation soil layer under all parameter combination conditions.

[0009] Among them, the obtaining of the measured data of typical projects includes burying 1 settlement observation board and 1 pressure monitoring device every 2 meters during the actual construction of the high-fill subgrade, continuously monitoring for 30 days, recording the settlement observation data of the high-fill subgrade, and comparing the settlement observation data of the high-fill subgrade with the theoretical prediction results.

[0010] Among them, the establishment of the interface element model for the filler and the foundation soil layer includes obtaining the friction angle parameter, the cohesion parameter, and the stiffness parameter at the interface between the filler and the foundation soil layer through an interface shear test, and the stiffness parameter at the interface between the filler and the foundation soil layer includes a normal stiffness parameter and a tangential stiffness parameter.

[0011] Among them, the use of the layered cumulative calculation method includes dividing the high-fill subgrade into 10 calculation units, calculating the stress increment value of each layer of calculation units based on the stress transfer law obtained from the parameter database of the filler and the foundation soil layer, and combining the compression index to determine the deformation amount of each layer of calculation units, and accumulating the deformation amounts of each layer of calculation units to obtain the total settlement.

[0012] Among them, the introduction of the time effect correction factor includes determining the consolidation parameters of the foundation soil layer through long-term consolidation tests. The consolidation parameters of the foundation soil layer include the consolidation coefficient and the degree of consolidation, and consider the characteristics of the settlement of the foundation soil layer changing with time under the dual effects of compression and consolidation.

[0013] Among them, the construction of the settlement intelligent prediction system includes inputting the parameter database of the filler and the foundation soil layer, the calibration coefficient, the deformation amount of each layer calculation unit, the time effect correction factor, and the prediction deviation value, and using the deep learning training algorithm to generate the settlement prediction model of the high embankment roadbed.

[0014] Among them, the evaluation test of the interface slip influence includes simulating the relative sliding phenomenon between the filler and the foundation soil layer by using an interface shear instrument, measuring the critical slip condition parameters between the filler and the foundation soil layer and the slip amount between the filler and the foundation soil layer. The critical slip condition parameters between the filler and the foundation soil layer include the critical normal stress and the critical shear stress; the critical slip condition parameters and the slip amount between the filler and the foundation soil layer are used to correct the prediction deviation value caused by ignoring the interface effect in the traditional settlement calculation method.

[0015] Among them, the layered summation calculation method is a calculation method that divides the high embankment roadbed and the foundation soil layer into several calculation units according to the soil layer properties, calculates the compression deformation of each calculation unit under the action of additional stress respectively, and then adds the deformation amounts of each calculation unit to obtain the total settlement amount.

[0016] The present invention systematically studies the interaction law between the filler and the foundation soil layer interface by establishing a high embankment roadbed interface test system, parameter change experiments and interface element models, and fundamentally solves the problem of insufficient settlement prediction accuracy in the prior art.

[0017] This method creatively constructs a 4×4 sensor monitoring network for the interface between the filler and the foundation soil layer, and obtains the first-hand experimental data of the stress distribution and deformation characteristics at the interface; through the parameter change experimental system, it systematically studies the influence of different parameter combinations on the interface mechanical properties, and establishes a comprehensive parameter database; introduces the concept of interface elements, regards the interface between the filler and the foundation soil layer as an independent mechanical unit, and accurately characterizes the interface friction characteristics and slip conditions through interface shear tests; at the same time, combined with the time effect correction factor and the evaluation of the interface slip influence, a settlement prediction model that takes into account both the interface effect and the time effect is established.

[0018] The method of the present invention overcomes the defect of ignoring the interface effect in the traditional technology, accurately characterizes the complex mechanical behavior of the interface between the filler and the foundation soil layer, and establishes a settlement calculation model considering the interface slip phenomenon. By combining laboratory research with engineering verification and introducing a deep learning algorithm to construct an intelligent prediction system, this method significantly improves the accuracy and reliability of the settlement prediction of high-fill subgrade, provides a scientific basis for engineering design, construction and monitoring, and effectively ensures the safety and stability of high-fill subgrade projects. Description of the Drawings

[0019] Figure 1 It is a flow chart of the method of the present invention. Detailed Embodiments

[0020] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0021] As Figure 1 shown, it is a flow chart of a method for predicting the settlement of a high-fill subgrade provided by the present invention. This method includes the following steps:

[0022] S01. Establish a high-fill subgrade interface test system, construct a large indoor test box with dimensions of 1000×500×500 mm, fill it with test fillers and simulate the foundation soil layer, and install 16 high-precision displacement sensors and 16 stress sensors at the contact interface between the filler and the foundation soil layer to form a 4×4 monitoring network, and measure the stress distribution values and deformation amounts at the interface under the self-weight of the filler.

[0023] S02. Conduct an experiment on changing test parameters. By changing the density index of the test filler, the water content index of the test filler, the gradation composition index of the test filler, and the bearing capacity index of the foundation soil layer, set the above indexes to 3 different values respectively, measure the friction coefficient parameter, the bond strength parameter and the shear parameter at the interface between the filler and the foundation soil layer under all parameter combination conditions, and establish a parameter database of the filler and the foundation soil layer.

[0024] S03. Obtain the measured data of 3 typical projects. During the actual construction of the high-fill subgrade, bury 1 settlement observation board and 1 pressure monitoring device every 2 meters, continuously monitor for 30 days, record the settlement observation data of the high-fill subgrade, compare the settlement observation data of the high-fill subgrade with the theoretical prediction results, and obtain the calibration coefficient.

[0025] S04. Establish an interface element model between the fill and the foundation soil layer. Consider the interface between the fill and the foundation soil layer as an independent mechanical unit. Obtain the friction angle parameter, cohesion parameter, and stiffness parameter of the interface between the fill and the foundation soil layer through interface shear tests. The stiffness parameter of the interface between the fill and the foundation soil layer includes the normal stiffness parameter and the tangential stiffness parameter.

[0026] S05. Adopt the layered cumulative calculation method. Divide the high embankment subgrade into 10 calculation units. Based on the stress transfer law obtained from the parameter database of the fill and the foundation soil layer, calculate the stress increment value of each layer of calculation units, and determine the deformation amount of each layer of calculation units in combination with the compression index. Accumulate the deformation amounts of each layer of calculation units to obtain the total settlement.

[0027] S06. Introduce a time effect correction factor. Determine the consolidation parameters of the foundation soil layer through long-term consolidation tests. The consolidation parameters of the foundation soil layer include the consolidation coefficient and the degree of consolidation. Consider the time-varying characteristics of the settlement of the foundation soil layer under the dual effects of compression and consolidation. Input the consolidation coefficient and the degree of consolidation, and calculate and plot the settlement time history prediction curve.

[0028] S07. Conduct an interface slip influence assessment test. Use an interface shear instrument to simulate the relative sliding phenomenon between the fill and the foundation soil layer, and measure the critical slip condition parameters and the slip amount between the fill and the foundation soil layer. The critical slip condition parameters between the fill and the foundation soil layer include the critical normal stress and the critical shear stress. Input the critical slip condition parameters and the slip amount between the fill and the foundation soil layer to correct the prediction deviation value caused by ignoring the interface effect in the traditional settlement calculation method.

[0029] S08. Construct a settlement intelligent prediction system. Input the parameter database of the fill and the foundation soil layer, the calibration coefficient, the deformation amount of each layer of calculation units, the time effect correction factor, and the prediction deviation value. Adopt a deep learning training algorithm to generate a settlement prediction model for the high embankment subgrade.

[0030] S09. Implement subgrade settlement monitoring and verification. Obtain the measured value of the subgrade settlement. Compare the measured value of the subgrade settlement with the calculation result of the settlement prediction model for the high embankment subgrade, calculate the prediction accuracy index and the reliability index, and output the settlement prediction report for the high embankment subgrade.

[0031] Among them, the interface test system for the high embankment subgrade specifically refers to a laboratory simulation system including a large test box, a simulated fill layer, a simulated foundation soil layer, a sensor network, and a data acquisition device, which is used to study the interaction law between the fill and the foundation soil layer.

[0032] Among them, the experimental study on the change of test parameters specifically refers to a systematic experimental study on the basis of the high-fill subgrade interface test system, which systematically changes the density index of the test filler, the water content index of the test filler, the gradation composition index of the test filler, and the bearing capacity index of the foundation soil layer, and studies the influence law of different parameter combinations on the settlement characteristics of the high-fill subgrade.

[0033] Among them, the calibration coefficient specifically refers to the proportional factor established by comparing the difference between the measured settlement data and the theoretical calculation results, which is used to correct the accuracy of the theoretical calculation and reflects the relationship between theory and practice.

[0034] Among them, the layered cumulative calculation method specifically refers to a calculation method in which the high-fill subgrade and the foundation soil layer are divided into several calculation units according to the soil layer properties, the compression deformation of each calculation unit under the action of additional stress is calculated respectively, and then the deformation amounts of each calculation unit are added to obtain the total settlement amount.

[0035] Among them, the interface shear test specifically refers to an experimental method using a direct shear instrument or a ring shear instrument to measure the shear strength parameters of the contact surface between the filler and the foundation soil layer under different normal pressures, which is used to determine the friction characteristics and slip conditions of the interface between the filler and the foundation soil layer.

[0036] Among them, the time effect correction factor specifically refers to the correction coefficient introduced considering the time-varying characteristics of the compression and consolidation of the foundation soil layer, which is used to adjust the settlement calculation results to reflect the time dependence of the settlement process.

[0037] Among them, the critical slip condition parameters of the filler and the foundation soil layer specifically refer to the critical stress state parameters required to satisfy when relative sliding occurs at the interface between the filler and the foundation soil layer. When the shear stress at the interface exceeds the critical value, obvious relative displacement will occur, affecting the overall settlement characteristics of the subgrade.

[0038] Among them, the prediction deviation value specifically refers to the calculation error caused by the traditional settlement calculation method ignoring the slip effect at the interface between the filler and the foundation soil layer, which is determined through the interface slip effect evaluation test and used to correct the calculation results.

[0039] Among them, the prediction accuracy index specifically refers to the deviation degree between the calculation result of the settlement prediction model and the measured value, usually expressed by the mean relative error and the root mean square error, which reflects the accuracy of the model prediction result.

[0040] Among them, the reliability index specifically refers to the ability evaluation of the settlement prediction model to maintain stable prediction accuracy under different working conditions, which is obtained through the statistical analysis of multiple groups of verification data and reflects the credibility of the model prediction result.

[0041] The following is a detailed description of the specific implementation manners of the above steps.

[0042] The specific implementation of step S01 is as follows: First, a rectangular test box with dimensions of 1000×500×500 mm is prepared in the laboratory. The box is made of high-strength steel, and the side wall thickness is not less than 15 mm to ensure no deformation during the experiment. A foundation soil layer with a thickness of not less than 200 mm is evenly laid at the bottom of the box, and the layered compaction method is used to ensure that the compactness of the foundation soil layer reaches more than 95%. Test fillers are placed on top of the foundation soil layer, and the filler thickness is not less than 250 mm. The layered compaction method is also used to ensure that the compactness of the fillers meets the design requirements. A 4×4 monitoring network is arranged at the contact interface between the fillers and the foundation soil layer, that is, 1 measuring point is arranged every 250 mm along the length direction of the test box, and 1 measuring point is arranged every 125 mm along the width direction. One high-precision displacement sensor and one high-precision stress sensor are installed at each measuring point. The accuracy of the displacement sensor is not less than 0.01 mm, and the accuracy of the stress sensor is not less than 0.1 kPa. The displacement sensor uses a differential transformer type displacement sensor, and the stress sensor uses a vibrating wire earth pressure cell. All sensors are connected to the computer terminal through a data acquisition device, and the acquisition frequency is set to 1 time per second, and continuous monitoring is carried out for not less than 24 hours to record the stress distribution value and deformation amount at the interface under the self-weight of the fillers. This step obtains the mechanical response data at the interface under the action of the self-weight of the fillers by constructing a high-fill subgrade interface test system, providing basic data support for subsequent analysis.

[0043] The specific implementation of step S02 is as follows: Based on the high fill subgrade interface test system, the test parameters are systematically changed for comparative experiments. First, set the test filler density index to three different levels, namely slightly compacted (relative density 85% - 90%), moderately compacted (relative density 90% - 95%), and heavily compacted (relative density 95% - 100%). Then set the test filler water content index to three different levels, namely on the dry side (optimum water content minus 3%), optimum water content state, and on the wet side (optimum water content plus 3%). Next, set the test filler gradation composition index to three different levels, namely a fine-grained ratio with about 30% fine-grained soil, a medium ratio with 15% - 20% fine-grained soil, and a coarse-grained ratio with less than 10% fine-grained soil. Finally, set the foundation soil layer bearing capacity index to three different levels, namely low bearing capacity (characteristic bearing capacity value below 80 kPa), medium bearing capacity (characteristic bearing capacity value 80 - 120 kPa), and high bearing capacity (characteristic bearing capacity value above 120 kPa). According to the orthogonal test design method, determine the test combination plan and conduct multiple groups of comparative tests. For each group of tests, measure the friction coefficient parameter, the bond strength parameter, and the shear parameter at the interface between the filler and the foundation soil layer. Among them, the friction coefficient parameter is obtained through a direct shear test, the bond strength parameter is obtained through a triaxial compression test, and the shear parameter is obtained through a ring shear test. Establish a parameter database for the filler and the foundation soil layer based on the test results, and use a four-dimensional interpolation algorithm to achieve the prediction of the interface mechanical properties under different parameter combinations. This step obtains the interface mechanical property parameters between the filler and the foundation soil layer under different working conditions through systematic parameter change tests, providing a basic database for subsequent model establishment.

[0044] The specific implementation of step S03 is as follows: Select three representative high embankment subgrade projects, which respectively represent high embankment subgrade engineering cases under soft soil foundation, medium-hard soil foundation and hard soil foundation conditions. Along the cross-section direction of the subgrade in each project, settlement observation sections are arranged. For each section, one settlement observation plate and one pressure monitoring device are buried every 2 meters along the width direction of the subgrade. The settlement observation plate is a steel circular plate with a diameter of not less than 300 mm and a thickness of not less than 10 mm, and a steel vertical rod with a diameter of not less than 20 mm is welded on the upper part; the pressure monitoring device is an earth pressure cell with a sensitivity of not less than 0.1 kPa. The elevation of the settlement observation plate is measured by a precision level, and the measurement accuracy is not less than 0.1 mm; the stress of the pressure monitoring device is monitored by a data acquisition instrument, and the monitoring frequency is 2 times a day. Continuously monitor for 30 days to obtain the settlement observation data and stress distribution data of the high embankment subgrade. At the same time, according to the filler parameters and foundation soil layer parameters, the theoretical settlement value is calculated by the layered summation method. Compare the measured settlement with the theoretical calculated settlement, and calculate the calibration coefficient K. The value of K is the ratio of the measured settlement to the theoretical settlement. The calibration coefficients of the three projects are respectively taken as K1, K2 and K3, and an interpolation table of calibration coefficients is established according to the type of foundation soil for settlement prediction correction under different foundation conditions. This step obtains the calibration coefficient by comparing the measured data with the theoretical calculation results, providing a correction basis for improving the prediction accuracy.

[0045] The specific implementation of step S04 is as follows: Establish an interface element model of the filler and the foundation soil layer by using the interface mechanics theory. The interface between the filler and the foundation soil layer is regarded as a zero-thickness interface element with independent mechanical properties, having independent strength parameters and deformation parameters. Obtain the interface friction angle parameter φ i , interface cohesion parameter c i through large-scale direct shear tests. In the test, a large direct shear apparatus with a diameter of not less than 300 mm is used, and shear tests are carried out under four different normal stress levels of 50 kPa, 100 kPa, 150 kPa and 200 kPa, and the shear stress-normal stress relationship curve is plotted. The interface friction angle parameter and the interface cohesion parameter are determined by the Coulomb strength criterion. At the same time, obtain the interface normal stiffness parameter k n and the interface tangential stiffness parameter k s。The interface stiffness test uses an interface test device to determine the interface stiffness parameters by measuring the normal and tangential deformations of the interface under different stress levels. The interface elements follow the Mohr-Coulomb strength criterion and the Coulomb friction law, and their constitutive relations are described by elastoplastic constitutive equations. This step provides a mechanical model basis for settlement calculation considering interface effects by establishing an interface element model between the fill and the foundation soil layer to obtain the mechanical property parameters of the interface. Among them, the interface test device is a special equipment for measuring the mechanical properties of the interface between the fill and the foundation soil layer. Common ones include large direct shear apparatuses, ring shear apparatuses, and triaxial interface apparatuses. For example, a large direct shear apparatus can measure the shear characteristics of the interface under different normal stresses to determine the interface stiffness parameters and provide basic data for the interface element model.

[0046] The specific implementation of step S05 is as follows: The layered summation method is used for the settlement calculation of the high embankment subgrade. First, the high embankment subgrade and the foundation soil layer are divided into 10 calculation units according to the soil layer properties, and the thickness of each calculation unit is determined according to the soil layer characteristics, generally not exceeding 3 meters. Based on the Boussinesq stress distribution theory, the additional stress increment value Δσ at the center point of each layer of the calculation unit is calculated. i , and the calculation formula adopts the stress superposition principle, considering the influence of the subgrade cross-sectional shape on the stress distribution. Combining the stress transfer law in the aforementioned parameter database of the fill and the foundation soil layer, the stress attenuation coefficient of the traditional Boussinesq solution is corrected, especially considering the stress mutation characteristics at the interface between the fill and the foundation soil layer. Then, according to the compression index C c or the compression modulus E s of each calculation unit, the deformation amount Δs of each layer of the calculation unit is calculated. i . For cohesive soil layers, the e-lgp compression curve is used to calculate the compression deformation of each layer; for sandy soil layers, the linear elastic model is used to calculate the compression deformation of each layer. The deformation amounts of each layer of the calculation unit are accumulated to obtain the total settlement amount. This step calculates the total settlement of the subgrade through the layered summation calculation method, considering the stress transfer law and the soil layer compression characteristics.

[0047] The specific implementation of step S06 is as follows: A time effect correction factor is introduced to consider the time effect during the settlement process. First, the consolidation parameters of the foundation soil layer, including the consolidation coefficient C v and the degree of consolidation U t , are determined through long-term consolidation tests. The standard consolidation apparatus is used for the consolidation test to measure the relationship between the consolidation deformation and time of the foundation soil under different load levels and draw the degree of consolidation-time curve. According to the consolidation theory, the degree of consolidation U at different times t is calculated using Taylor's formula. t . Considering that in actual engineering, the settlement process is affected not only by consolidation but also by factors such as fill creep and secondary consolidation, a time effect correction factor α t。The time effect correction factor is determined by comparing the measured settlement time curve with the theoretical consolidation time curve, and the value of α t generally varies non-linearly with time and can be expressed in the form of a power function: α t = a·t b , where a and b are fitting parameters determined by regression analysis of the measured data. Finally, the settlement prediction formula considering the time effect is obtained: S t = S·U t ·α t , where S t is the settlement at time t, S is the final settlement, U t is the degree of consolidation at time t, and α t is the time effect correction factor at time t. By introducing the time effect correction factor in this step, the time-dependent characteristics of the settlement process are considered to achieve the prediction of the settlement time history.

[0048] The specific implementation method of step S07 is as follows: conduct an interface slip influence evaluation test to study the influence of the interface slip between the filler and the foundation soil layer on the settlement calculation. Use a self-developed large-scale interface shear apparatus to simulate the interface slip process between the filler and the foundation soil layer. The interface shear apparatus consists of an upper shear box, a lower shear box, a normal loading system, a horizontal loading system, and a displacement measurement system. The upper shear box is filled with test filler, and the lower shear box is filled with foundation soil, forming an artificial interface between the upper and lower shear boxes. By applying different levels of normal loads and shear loads, measure the stress-displacement relationship curve at the interface to determine the critical slip condition parameters of the filler and the foundation soil layer, including the critical normal stress σ c and the critical shear stress τ c . When the stress state at the interface satisfies τ>τ c or σ<σ c , obvious relative slip will occur at the interface, resulting in deviation in the settlement calculation of the subgrade. Measure the interface slip amount δ s under different stress states through the test, and establish a relationship model between the interface slip amount and the stress state. Convert the interface slip amount into an equivalent settlement amount to obtain the predicted deviation value ΔS s caused by the interface slip. This predicted deviation value is usually related to the filler height H, the filler density ρ f , and the foundation bearing capacity f k , and can be expressed as ΔS s = β·H·ρ f / f k , where β is a proportionality coefficient determined by regression analysis of the test data. By conducting the interface slip influence evaluation test in this step, study the influence of the interface slip on the settlement calculation, and provide a correction basis for improving the settlement prediction accuracy.

[0049] The specific implementation of step S08 is as follows: Construct a settlement intelligent prediction system and establish a settlement prediction model for high-fill subgrade using deep learning methods. First, construct a deep neural network model framework with a multi-layer perceptron structure, including an input layer, three hidden layers, and an output layer. The number of nodes in the input layer is 20, corresponding to various parameter indicators of the filler and foundation soil layers; the number of nodes in the first hidden layer is 32, and the ReLU function is used as the activation function; the number of nodes in the second hidden layer is 16, and the ReLU function is used as the activation function; the number of nodes in the third hidden layer is 8, and the ReLU function is used as the activation function; the number of nodes in the output layer is 2, corresponding to the predicted settlement amount and the parameters of the settlement time curve respectively. Use the data in the filler and foundation soil layer parameter database as training samples, and the input parameters include filler density, filler water content, filler gradation composition, foundation bearing capacity, etc.; the output labels are the settlement amount and the parameters of the time curve under each working condition. Train the neural network using the backpropagation algorithm, use the mean square error function as the loss function, use the Adam optimizer as the optimization algorithm, set the learning rate to 0.001, and set the number of training rounds to 1000 rounds. Use parameters such as the calibration coefficient, the deformation amount of each layer of computing units, the time effect correction factor, and the prediction deviation value as the auxiliary input of the neural network to improve the prediction accuracy of the model. After training, use the cross-validation method to evaluate the model performance, and the accuracy of the validation set is not less than 90%. This step constructs a deep learning model, comprehensively considers various influencing factors, and establishes a high-precision settlement prediction model for high-fill subgrade.

[0050] The specific implementation of step S09 is as follows: Implement subgrade settlement monitoring and verification to evaluate the accuracy and reliability of the settlement prediction model. Select 3 newly built high-fill subgrade projects as verification projects and install a settlement monitoring system, including monitoring equipment such as settlement observation plates, multi-point displacement gauges, and earth pressure cells. One settlement monitoring section is set every 50 meters along the subgrade cross-section, and no less than 5 monitoring points are arranged along the subgrade width direction for each section. The monitoring frequency is 1 time per day in the initial stage (within 30 days after filling), 1 time every 3 days in the middle stage (30 - 90 days after filling), and 1 time per week in the later stage (more than 90 days after filling), and continuous monitoring is carried out for no less than 180 days. Obtain the measured subgrade settlement value S m , and at the same time calculate the theoretical settlement value S c using the aforementioned high-fill subgrade settlement prediction model. Compare the measured value with the calculated value, and calculate the prediction accuracy indicators, including the mean relative error and the root mean square error where n is the number of monitoring points. At the same time, calculate the reliability index That is, the probability that the predicted relative error does not exceed 15%. When the average relative error does not exceed 10% and the reliability index is not less than 85%, the prediction model is considered to have high accuracy. According to the verification results, a settlement prediction report for the high-fill subgrade is output, including the predicted settlement amount, settlement time curve, settlement contour map, settlement risk assessment, etc. This step evaluates the accuracy and reliability of the settlement prediction model through the implementation of subgrade settlement monitoring and verification, and forms a complete technical system for the high-fill subgrade settlement prediction method.

[0051] Among them, the establishment of the interface element model in step S04 is the key innovation point for the high-fill subgrade settlement prediction. By introducing independent interface elements, the problem that the traditional settlement calculation method ignores the interface effect is solved. The constitutive relationship of the interface element can be expressed as:

[0052]

[0053] Where Δσ n is the interface normal stress increment, Δτ is the interface shear stress increment, Δu n is the interface normal displacement increment, Δu s is the interface shear displacement increment, k n is the interface normal stiffness, k s is the interface shear stiffness. The strength criterion of the interface element satisfies the Mohr-Coulomb criterion:

[0054] τ ≤ c i + σ n tan φ i ;

[0055] Where τ is the interface shear stress, σ n is the interface normal stress, c i is the interface cohesion, φ i is the interface friction angle. When the shear stress reaches the limit value of the strength criterion, the interface will slip and enter the plastic state.

[0056] The layered cumulative calculation method in step S05 is based on the combined application of the Boussinesq stress distribution theory and the soil compression theory. For a rectangular uniformly distributed load, the vertical additional stress at the center point of the unit can be expressed as:

[0057] Δσ z = q·I z ;

[0058] Where q is the additional stress generated by the self-weight of the filler, I z is the stress influence coefficient, and its calculation formula is:

[0059]

[0060] where x and y are the horizontal distances from the calculation point to the load center, and z is the vertical distance from the calculation point to the load acting surface. Considering the actual trapezoidal shape of the subgrade, the principle of stress superposition needs to be used for calculation.

[0061] For the cohesive soil layer, the calculation formula for its compression deformation is:

[0062]

[0063] where h i is the thickness of the i-th soil layer, e0 is the initial void ratio, C c is the compression index, p0 is the initial effective stress, and Δσ i is the additional stress increment.

[0064] For the sandy soil layer, the calculation formula for its compression deformation is:

[0065]

[0066] where E s is the compression modulus of the soil layer.

[0067] The time effect correction factor in step S06 takes into account the difference between the consolidation theory and the actual settlement process. According to the one-dimensional consolidation theory, the consolidation degree calculation formula for the foundation soil layer is:

[0068]

[0069] where T v is the time factor, C v is the consolidation coefficient, t is the consolidation time, and H d is the drainage path length.

[0070] In engineering practice, the approximate calculation formula is usually adopted:

[0071]

[0072] The introduction of the time effect correction factor α t takes into account the non-linear characteristics in the actual settlement process. In its power function expression α t = a·t b , the parameters a and b are determined by regression analysis of measured data. Generally, the value range of a is 0.8 - 1.2, and the value range of b is -0.2 - 0.2.

[0073] The interface slip influence evaluation in step S07 is another innovation point of this method. The relationship between the interface slip amount and the stress state can be expressed as:

[0074] (when τ > τ c )

[0075] Where λ is the slip coefficient, determined by experiments, and generally ranges from 0.01 to 0.1 mm / kPa; v is the interfacial shear stress, τ c is the critical shear stress, σ n is the interfacial normal stress.

[0076] The predicted deviation value ΔS caused by interfacial slip s is related to the filler height H, filler density ρ f , and the foundation bearing capacity f k as follows:

[0077] ΔS s = β·H·ρ f / f k ;

[0078] Where β is the proportionality coefficient, determined by regression analysis of experimental data, and generally ranges from 0.001 to 0.01.

[0079] The deep learning model in step S08 adopts a multi-layer perceptron structure, and the output calculation formula of each neuron is:

[0080] y = f(w T x + b);

[0081] Where f is the activation function, w is the weight vector, x is the input vector, and b is the bias term. The expression of the ReLU activation function is:

[0082] f(x) = max(0, x);

[0083] The loss function adopts the mean square error function:

[0084]

[0085] Where y i is the actual settlement, is the predicted settlement, and N is the number of samples.

[0086] The parameter update rule of the Adam optimizer is:

[0087] m t = β1·m t-1 + (1 - β1)·g t ;

[0088]

[0089]

[0090] Where g t is the current gradient, m tis the first-order momentum, v t is the second-order momentum, and is the corrected momentum, η is the learning rate, β1, β2 and ∈ are hyperparameters, generally taking β1 = 0.9, β2 = 0.999, ∈ = 10 -8 .

[0091] Specifically, the principle of the present invention is: The core technical principle of the present invention lies in systematically characterizing the interaction mechanism between the filler and the foundation soil layer interface, and establishing a settlement prediction model for high embankment subgrades considering the interface effect. Traditional settlement prediction methods usually regard the filler and the foundation as a continuum, ignoring the special mechanical behavior at the interface, while the present invention regards the interface as an independent mechanical unit, and through the combination of experiments and theories, comprehensively reveals the key influence of the interface effect on settlement prediction.

[0092] First of all, the present invention constructs a large indoor test box with dimensions of 1000×500×500 mm to simulate the high embankment subgrade structure, installs high-precision displacement sensors and stress sensors at the contact interface between the filler and the foundation soil layer, and forms a 4×4 monitoring network. This design enables the stress distribution and deformation characteristics at the interface to be accurately captured. By changing parameters such as the filler density, water content, gradation composition, and foundation bearing capacity, the variation laws of the interface mechanical parameters under different working conditions are systematically studied, and a comprehensive parameter database is established, providing a reliable data basis for the subsequent prediction model.

[0093] In terms of constructing the interface mechanical model, the present invention determines the interface friction angle, cohesion, and stiffness parameters through interface shear tests, accurately characterizing the interface mechanical properties. The layered cumulative calculation method is used to deal with the settlement problem of high embankment subgrades, avoiding the errors caused by the traditional continuous medium hypothesis. A time effect correction factor is introduced to consider the variation law of settlement with time under the dual effects of compression and consolidation. In particular, through the interface slip influence evaluation test, the present invention systematically studies the relative sliding phenomenon between the filler and the foundation soil layer, determines the critical slip condition parameters and the slip amount, and corrects the prediction deviation caused by the traditional method ignoring the interface slip effect.

[0094] In addition, the present invention combines deep learning algorithms, and based on a large amount of experimental data and engineering measured data, constructs an intelligent prediction system, realizing an effective transformation from laboratory research to engineering application. Through actual engineering verification, this method significantly improves the prediction accuracy and solves the problem of insufficient prediction accuracy of high embankment subgrades that is difficult to solve by traditional methods.

[0095] A specific Embodiment 1 of the present invention is provided below. In a certain highway construction project, the route passes through a hilly area, and a high fill subgrade with a height of up to 23 meters needs to be built. The lower part of the subgrade is a soft clay foundation, and the stability and settlement control of the subgrade have become key problems in the project construction. The construction personnel used the settlement prediction method for high fill subgrade of the present invention for settlement analysis and control, and the specific implementation process is as follows.

[0096] First, the construction personnel established a high fill subgrade interface test system. An indoor large test box with dimensions of 1000×500×500 mm was constructed, with 200 mm thick foundation soil filled at the bottom and 300 mm thick subgrade filler filled at the upper part. 16 high-precision displacement sensors and 16 stress sensors were arranged at the contact interface between the filler and the foundation soil to form a 4×4 monitoring network. The accuracy of the displacement sensor is 0.01 mm, and the accuracy of the stress sensor is 0.1 kPa. Continuous monitoring was carried out for 48 hours, and the stress distribution and deformation amount at the interface under the self-weight of the filler were recorded. The monitoring results show that the interface stress distribution is non-uniform, with larger stress in the central area and smaller stress in the edge area. The average stress value is 42.3 kPa, and the maximum deformation amount is 0.87 mm.

[0097] Subsequently, the construction personnel carried out an experiment on changing test parameters. Three filler density grades, three water content levels, and three filler gradation compositions were set, and the specific parameters are shown in Table 1:

[0098] Table 1 Setting Table of Filler Parameters

[0099] Parameter type Level 1 Level 2 Level 3 Packing density Light compaction (85%) Medium compaction (92%) Heavy compaction (98%) Water content On the dry side (12%) Optimal (16%) On the wet side (20%) Gradation composition Fine-grained soil (30%) Medium (18%) Coarse-grained (8%)

[0100] At the same time, three bearing capacity grades were also set for the foundation soil layer: low bearing capacity (65 kPa), medium bearing capacity (105 kPa), and high bearing capacity (135 kPa). Using the orthogonal test method, the construction personnel determined 9 groups of key test combinations. Through direct shear tests, triaxial compression tests, and ring shear tests, the friction coefficient, cohesion strength, and shear parameters at the interface between the filler and the foundation soil layer under different combination conditions were measured. The test results show that the filler density and the foundation bearing capacity have the most significant influence on the interface friction characteristics, and the correlation coefficients reach 0.76 and 0.83 respectively.

[0101] Next, the construction workers selected 3 typical engineering cases, which respectively represent the high embankment subgrade projects under soft soil foundation, medium-hard soil foundation and hard soil foundation conditions. Settlement observation sections were arranged along the roadbed cross-section at each location, and 1 settlement observation board and 1 pressure monitoring device were buried every 2 meters along the roadbed width direction for each section. Continuous monitoring was carried out for 30 days, and the settlement observation data of the high embankment subgrade were recorded. By comparing the measured settlement with the theoretically calculated settlement, the calibration coefficient was calculated. The calibration coefficients under the three foundation conditions are: 1.23 for soft soil foundation, 1.08 for medium-hard soil foundation and 0.96 for hard soil foundation.

[0102] Subsequently, the construction workers established an interface element model between the filler and the foundation soil layer. The interface friction angle parameter φ i and the interface cohesion parameter c i were obtained through large-scale direct shear tests. Shear tests were carried out under four different normal stress levels of 50 kPa, 100 kPa, 150 kPa and 200 kPa, and the interface mechanical parameters under different working conditions were obtained, as shown in Table 2:

[0103] Table 2 Interface Mechanical Parameter Table

[0104]

[0105] Using the layered cumulative calculation method, the construction workers divided the high embankment subgrade into 10 calculation units, and the thickness of each layer was 2.3 meters. Based on the Boussinesq stress distribution theory and the filler and foundation soil layer parameter database of the present invention, the additional stress increment values at the center points of each layer of calculation units were calculated. For the soft soil foundation, the stress increment values of each layer of calculation units were 183.6, 167.2, 152.8, 139.5, 127.3, 116.0, 105.6, 96.0, 87.2 and 79.2 kPa in sequence. Combining the compression index C c and the compression modulus E s of each layer of soil, the deformation amounts of each layer of calculation units were calculated, and the total settlement was obtained as 286 mm by accumulation.

[0106] Subsequently, the construction workers introduced a time effect correction factor. The consolidation coefficient C v of the foundation soil layer was determined through long-term consolidation tests to be 3.5×10 -7 m 2 / s and the degree of consolidation U t at different time periods. Considering the time-varying characteristics of the settlement of the foundation soil layer under the dual action of compression and consolidation, the time effect correction factor α t =1.15×t -0.12 was introduced, where t is the number of days. The settlement time history prediction curve was calculated and plotted, and the cumulative settlement amounts at 100 days, 200 days and 365 days were predicted to be 196 mm, 238 mm and 267 mm respectively.

[0107] Next, the construction workers carried out an interface slip influence assessment test. Through a self-developed large-scale interface shear instrument, the influence of the interface slip between the filler and the foundation soil layer on settlement calculation was studied. The critical slip condition parameters were measured, including a critical normal stress of 75 kPa and a critical shear stress of 38 kPa. When the stress state at the interface satisfies that the shear stress is greater than 38 kPa or the normal stress is less than 75 kPa, obvious relative slip will occur at the interface. The experimentally measured slip coefficient λ was 0.05 mm / kPa, resulting in a settlement calculation deviation value of approximately 27 mm.

[0108] Based on the above experimental and calculation results, the construction workers constructed a settlement intelligent prediction system. A deep neural network model with a multi-layer perceptron structure was adopted. The number of nodes in the input layer was 20, the number of nodes in the three hidden layers was 32, 16, and 8 respectively, and the number of nodes in the output layer was 2. The input parameters included filler density, water content, gradation composition, and foundation bearing capacity, etc.; the output was the predicted settlement amount and the parameters of the settlement time curve. Using the backpropagation algorithm, the loss function was the mean square error function, and the learning rate was 0.001. After training for 1000 rounds, a settlement prediction model for high embankment subgrades was obtained. The prediction accuracy of the model on the validation set reached 92.7%.

[0109] Finally, the construction workers carried out subgrade settlement monitoring and verification at another high embankment subgrade section of this expressway. A settlement monitoring system was arranged, including monitoring equipment such as settlement observation plates, multi-point displacement gauges, and earth pressure cells. One monitoring section was set every 50 m along the subgrade cross-section, and 5 monitoring points were arranged along the subgrade width direction for each section. The monitoring results showed that the measured settlement amounts on the 30th day, 90th day, and 180th day after filling were 112 mm, 187 mm, and 226 mm respectively, while the predicted settlement amounts were 106 mm, 178 mm, and 217 mm respectively. The predicted accuracy indexes were calculated, with an average relative error of 6.2%, a root mean square error of 8.7 mm, and a reliability index of 91.3%.

[0110] Traditional high embankment subgrade settlement prediction methods mainly rely on one-dimensional consolidation theory and empirical formulas, ignoring the influence of the interface slip effect and time effect between the filler and the foundation soil layer, resulting in a large deviation between the prediction results and the actual settlement. The average error is usually between 15% and 25%. Previous related studies generally adopted a simplified homogeneous foundation assumption, did not consider the three-dimensional distribution characteristics of stress, and did not fully study the interface mechanical properties under different parameter combinations. The applicability of these methods under complex geological conditions is poor, and the prediction accuracy is not high, making it difficult to meet the engineering requirements for the settlement control of high embankment subgrades on expressways.

[0111] It should be noted that the detailed explanations of the variables involved in the present invention are shown in Tables 3 and 4 below.

[0112] Table 3 Variable Explanation Table (First Part)

[0113]

[0114]

[0115] Table 4 Variable Explanation Table (Second Part)

[0116]

[0117] As described above, it is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention.

Claims

1. A method for predicting the settlement of a high fill subgrade, characterized in that, Including: Establish a high embankment subgrade interface test system, install displacement sensors and stress sensors at the interface between the filler and the foundation soil layer to form a monitoring network; execute test parameter change experiments, and establish a parameter database for the filler and the foundation soil layer; obtain measured data of typical projects to obtain calibration coefficients; establish an interface element model for the filler and the foundation soil layer to obtain interface mechanical parameters; use the layered accumulation calculation method to calculate the total settlement; introduce a time effect correction factor to calculate the settlement time history prediction curve; conduct an interface slip influence evaluation test to correct the prediction deviation value caused by ignoring the interface effect in the traditional settlement calculation method; construct a settlement intelligent prediction system to generate a high embankment subgrade settlement prediction model; implement subgrade settlement monitoring and verification, and output a high embankment subgrade settlement prediction report.

2. The method for predicting the settlement of a high embankment subgrade according to claim 1, characterized in that, The establishment of the high embankment subgrade interface test system includes constructing a large indoor test box with dimensions of 1000×500×500 mm, filling test fillers and simulating the foundation soil layer, installing 16 high-precision displacement sensors and 16 stress sensors at the contact interface between the filler and the foundation soil layer to form a 4×4 monitoring network, and measuring the stress distribution value and deformation amount at the interface under the self-weight of the filler.

3. The high embankment subgrade settlement prediction method according to claim 2, characterized in that, The execution of the test parameter change experiment includes changing the density index of the test filler, the water content index of the test filler, the gradation composition index of the test filler, and the bearing capacity index of the foundation soil layer, setting the above indexes to 3 different values respectively, and measuring the friction coefficient parameter, the bond strength parameter and the shear parameter at the interface between the filler and the foundation soil layer under all parameter combination conditions.

4. The method for predicting settlement of high embankment subgrade according to claim 3, characterized in that The obtaining of the measured data of typical projects includes burying 1 settlement observation board and 1 pressure monitoring device every 2 meters during the actual construction of the high embankment subgrade, continuously monitoring for 30 days, recording the settlement observation data of the high embankment subgrade, and comparing the settlement observation data of the high embankment subgrade with the theoretical prediction results.

5. The method for predicting the settlement of a high-fill subgrade according to claim 4, characterized in that The establishment of the interface element model for the filler and the foundation soil layer includes obtaining the friction angle parameter, the cohesion parameter and the stiffness parameter at the interface between the filler and the foundation soil layer through an interface shear test, and the stiffness parameter at the interface between the filler and the foundation soil layer includes the normal stiffness parameter and the tangential stiffness parameter.

6. The method for predicting the settlement of a high-fill subgrade according to claim 5, wherein The adoption of the layered accumulation calculation method includes dividing the high embankment subgrade into 10 calculation units, calculating the stress increment value of each layer of calculation unit based on the stress transfer law obtained from the parameter database of the filler and the foundation soil layer, and determining the deformation amount of each layer of calculation unit in combination with the compression index, and accumulating the deformation amounts of each layer of calculation unit to obtain the total settlement.

7. The method for predicting the settlement of a high embankment subgrade according to claim 6, wherein, The introduction of the time effect correction factor includes determining the consolidation parameters of the foundation soil layer through a long-term consolidation test, and the consolidation parameters of the foundation soil layer include the consolidation coefficient and the degree of consolidation, considering the characteristics of the settlement of the foundation soil layer changing with time under the dual effects of compression and consolidation.

8. The method for predicting the settlement of high-fill subgrade according to claim 7, characterized in that, The construction of the settlement intelligent prediction system includes inputting the parameter database of the filler and the foundation soil layer, the calibration coefficient, the deformation amount of each layer of calculation unit, the time effect correction factor, and the prediction deviation value, and using a deep learning training algorithm to generate a high embankment subgrade settlement prediction model.

9. The method for predicting the settlement of a high-fill subgrade according to claim 8, wherein The interface slip influence evaluation test is carried out, including simulating the relative sliding phenomenon between the filler and the foundation soil layer by using an interface shear instrument, measuring the critical slip condition parameters between the filler and the foundation soil layer and the slip amount between the filler and the foundation soil layer. The critical slip condition parameters between the filler and the foundation soil layer include the critical normal stress and the critical shear stress; the critical slip condition parameters and the slip amount between the filler and the foundation soil layer are used to correct the prediction deviation value caused by ignoring the interface effect in the traditional settlement calculation method.

10. The method for predicting settlement of high embankment subgrade according to claim 9, characterized in that, The layer-by-layer cumulative calculation method is a calculation method that divides the high-fill subgrade and the foundation soil layer into several calculation units according to the soil layer properties, calculates the compression deformation of each calculation unit under the action of additional stress respectively, and then adds the deformation amounts of each calculation unit to obtain the total settlement amount.

Images