Method for detecting vertical compression bearing capacity of foundation pile with expanded head

Abstract

The invention belongs to the technical field of foundation pile bearing capacity testing, and particularly relates to a method for measuring the vertical compressive bearing capacity of an expanded-head foundation pile, which comprises the following steps: S1, measuring to obtain measured data; s2, a load loading model is established, and pile bottom displacement zero clearing is completed; s3, drawing a prediction Q-S curve; s4, pile top settlement corresponding to each stage of load is obtained in a field test, and an actual Q-S curve is drawn; and S5, calculating a soil body influence coefficient according to the deviation amount between the predicted Q-S curve and the actual Q-S curve, correcting pile side resistance and pile end resistance according to the soil body influence coefficient, and calculating the vertical bearing capacity by using the corrected pile side resistance and pile end resistance. The calculated bearing capacity of the foundation pile is corrected according to the influence coefficient generated by the environment in the actual testing process, so that the systematic deviation between the limit load and the actual failure load of theoretical calculation is reduced, and the accuracy of calculation of the bearing capacity of the foundation pile is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of pile foundation bearing capacity testing, and in particular relates to a method for testing the vertical compressive bearing capacity of a pile foundation with an enlarged head. Background Art

[0002] When the bearing capacity of the enlarged head foundation pile is tested through a static load test, the load box is buried at the bottom of the pile, and the internal force of the pile is tested by placing a rebar meter at the pile body and the displacement test tube is pre-buried at the load surface change point of the pile body and at the bottom of the pile to complete the vertical compressive bearing capacity test of the single pile with an enlarged head, so as to obtain the vertical compressive bearing capacity of the single pile.

[0003] However, during actual testing, the uncertainty of the pile design conditions and the complexity of the testing environment can lead to unforeseen risks and challenges. First, from a design perspective, the design of enlarged-head piles relies on a precise understanding of soil parameters, geological conditions, and construction techniques. Furthermore, during actual testing, soil parameters such as ultimate resistance often exhibit significant variability. This uncertainty can lead to deviations in the design-phase estimation of the pile's bearing capacity.

[0004] Specifically, the calculation of the ultimate load depends on the empirical values of soil parameters and the idealization of pile materials, such as the soil shear strength formula Calculate the bearing capacity of foundation piles or generate graded loading schemes based on a simplified model of pile-soil interface stiffness. However, in actual engineering, soil has significant spatial variability (such as weak interlayers, pore water pressure changes) and nonlinear mechanical response, and the pile structure (such as enlarged head size deviation, concrete creep effect) also introduces uncertainty. These factors lead to a systematic deviation between the theoretically calculated ultimate load and the actual failure load, which in turn causes graded loading compensation. If the settings are inaccurate, underloading (failure to trigger true failure) or overloading (premature termination of loading) may easily occur during the actual test process, which will eventually cause the bearing capacity determination error to exceed the threshold allowed by the test (such as ±10%), affecting the safety and economy of the project. Summary of the Invention

[0005] In order to solve the above-mentioned problems existing in the prior art, the present invention provides a method for detecting the vertical compressive bearing capacity of piles with enlarged heads, which solves the problem that in the actual testing process, there is a systematic deviation between the theoretically calculated ultimate load and the actual destructive load, which in turn causes the error in the pile body bearing capacity determination to exceed the allowable threshold.

[0006] The object of the present invention can be achieved by the following technical solution: A method for detecting the vertical compressive bearing capacity of a pile with an enlarged head comprises the following steps: S1: Obtaining measurement data, the measurement data including soil layer parameters, pile structure parameters, and contact surface parameters, and constructing a three-dimensional pile-soil layer numerical model based on the measurement data; S2: Establish the load loading model and complete the pile bottom displacement reset; S3: Use the theoretically set load value and loading level to apply load step by step, obtain the pile top settlement corresponding to each load level, and draw the predicted QS curve; S4: Obtain the pile top settlement corresponding to each load level in the field test and draw the actual QS curve; S5: Calculating a soil influence coefficient based on the deviation between the predicted QS curve and the actual QS curve, correcting the pile side resistance and pile end resistance based on the soil influence coefficient, and calculating the vertical bearing capacity using the corrected pile side resistance and pile end resistance.

[0007] Preferably, in step S1, the soil layer parameters include the density, Poisson's ratio, cohesion, internal friction angle, elastic modulus and stiffness of the rock layer and soil layer at the pile implantation depth of the test point; The pile body structural parameters include the length and cross-sectional area of the pile body with a constant cross section, and the height and shape parameters of the pile body with a variable cross section.

[0008] Preferably, step S5 includes the following sub-steps: S51: Compare the predicted settlement corresponding to the predicted QS curve and the actual QS curve step by step and actual settlement , calculate the root mean square error; , n is the total number of load stages; S52: Calculate soil influence coefficient: , is the settlement threshold allowed by the test specification; S53: Calculate and correct pile side wear resistance and corrected pile tip resistance : , ,in, is the side wear resistance of the original pile, is the original pile end resistance; S54: Get the corrected vertical bearing capacity The calculation formula is: .

[0009] Preferably, step S5 further includes S6: using LSTM network to predict the pile top settlement trend, and in the predicted value When , an early warning is triggered, m is the predicted future n-th level load, where the training set of the LSTM network is the point element of the actual QS curve, and the validation set is the point element of the predicted QS curve.

[0010] Preferably, the input features of the LSTM network include point elements of the actual QS curve, calculated soil parameters and calculated pile top settlement acceleration, and the output is the settlement prediction value of the future n-level load.

[0011] Preferably, the training of the LSTM network includes: Data processing: the load value Q and pile top settlement value in the actual test are standardized using the Z-score standardization method; The actual QS curve is numerically differentiated to calculate the pile top settlement acceleration; The standardized data are combined in the order of the time series to form the input sample data set for LSTM network training; Build and train the LSTM network and define the mean square error function as the loss function; The training set data is input into the constructed LSTM network, and multiple rounds of iterative training are performed according to the set training parameters. The training parameters are the point set of the actual QS curve. The weight parameters of the network are then adjusted through back propagation of the optimization algorithm to minimize the loss function. The actual test data collected in real time is processed according to the above preprocessing method and input into the trained LSTM network for prediction.

[0012] Preferably, in step S1, the contact surface between the soil layer and the pile body is set using the Coulomb friction model, and the friction coefficient , is the friction angle between the pile and soil, which is determined through interface tests.

[0013] Preferably, the determination of the Coulomb model includes: Install the sample: fix the pile sample vertically on the base of the shear box, and put the soil sample on the outside of the pile to ensure that the contact surface is tightly fitted; Gradual application of normal stress: according to the actual soil distribution of the project, set multiple groups of normal stress ; Calculate the interface shear strength using the formula: , A is the contact area between the pile and the soil, draw , fitting the Coulomb formula by linear regression: , The slope of , the intercept is the soil cohesion c.

[0014] The beneficial effects of the present invention are: The present invention calculates the pile bearing capacity by using the corrected pile side resistance and pile end resistance, thereby correcting the calculated pile bearing capacity according to the influence coefficient generated by the environment during the actual test process, thereby reducing the systematic deviation between the theoretically calculated ultimate load and the actual failure load, and improving the accuracy of the pile bearing capacity calculation; Moreover, by utilizing the time series prediction capability of the LSTM network, combined with field test data and previously calculated soil parameters, an accurate prediction of the pile top settlement trend of piles with enlarged heads during subsequent loading is achieved, and early warnings can be triggered in a timely and effective manner, providing strong support for pile construction quality control and engineering safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] To facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0016] Figure 1 This is a structural block diagram of the detection method of the present invention. DETAILED DESCRIPTION

[0017] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0018] See also Figure 1 This embodiment provides a method for detecting the vertical compressive bearing capacity of a pile with an enlarged head, comprising the following steps: S1: obtaining measurement data by measurement, wherein the measurement data includes soil layer parameters, pile structure parameters, and contact surface parameters; Using geological drilling technology, multiple boreholes were placed at various locations around the foundation piles, reaching a depth below the projected pile tip to ensure coverage of all soil layers that could affect the pile's bearing properties. Untouched soil samples were removed from the boreholes and subjected to various physical and mechanical property tests using geotechnical testing instruments. Supplementary soil testing was also performed using in-situ testing methods such as the standard penetration test (SPT) and cone penetration test (CPT). The SPT involved hammering at varying depths, with the number of blows per 30 cm recorded. This further determined the compactness and mechanical property changes of the soil layer. The curves of cone tip resistance and sidewall friction resistance versus depth obtained from the cone penetration test were cross-correlated with the geotechnical test results to comprehensively determine the spatial distribution of soil parameters.

[0019] Soil parameters include density, Poisson's ratio, cohesion, internal friction angle, elastic modulus and stiffness of the rock layer and soil layer at the pile embedding depth of the test point; Detailed measurements of the actual dimensions of the foundation piles, including the length and cross-sectional area of the uniform-section pile, as well as the height and shape parameters of the variable-section pile. Non-destructive testing techniques, such as the low-strain reflection wave method, are then used to test the integrity of the pile concrete to determine if there are any defects in the pile and the elastic modulus of the concrete. At the interface between the pile and the soil, the mechanical properties of the contact surface significantly affect the bearing capacity of the pile due to the significant differences in the properties of the soil and pile materials. Indoor direct shear tests are conducted to simulate the shear behavior of the pile-soil interface. Samples with materials similar to those in the field, such as soil and pile, are prepared and subjected to shearing under different normal stresses to determine the friction coefficient μ of the pile-soil interface. The average value of multiple tests is used to determine the friction coefficient μ of the pile-soil interface. Constructing a 3D pile-soil numerical model: Based on the soil layer parameters, pile structure parameters, and contact surface parameters measured above, professional geotechnical engineering numerical simulation software, such as FLAC3D or ABAQUS, is used to construct a three-dimensional model of the pile-soil layer. In the model, the soil layer is simulated using solid elements, and the corresponding material properties are set according to the actual distribution and parameters of the different soil layers. The pile body is also simulated using solid elements, and the pile body concrete and the enlarged head are given their own mechanical parameters. Contact elements are set at the pile-soil interface to define the friction characteristics of the contact surface, so that the model can accurately simulate the interaction between the pile and the soil. The boundary conditions of the model are set as follows: the bottom boundary is a fixed constraint, which limits vertical and horizontal displacements; the side boundary is set as a horizontal constraint, allowing vertical displacement to simulate a semi-infinite soil space.

[0020] S2: Establish the load loading model and complete the pile bottom displacement reset: First, create a load loading model: In numerical simulation software, a corresponding load loading pattern is established based on the actual vertical load conditions that the foundation piles may bear. In this embodiment, the combination of vertical dead load and live load transmitted from the building superstructure is considered, and the total load value range is set and distributed according to a certain loading level; Then complete the pile bottom displacement reset: Before loading begins, the displacement of the pile bottom node in the model is initialized and its vertical displacement value is set to zero. This is to accurately capture the displacement changes caused by soil deformation and pile compression during the subsequent loading process, ensuring the accuracy of the calculation results.

[0021] S3: Draw the predicted QS curve: S301: Apply loads step by step using the theoretically set load values and loading levels to obtain the pile top settlement value corresponding to each load level: Start the numerical simulation calculation and apply vertical loads to the pile top step by step from small to large according to the theoretically set load value and loading level. After each load level is applied, the software solves the equilibrium equation of the soil and pile body and calculates the corresponding deformation state until the set maximum load is reached or the model convergence condition is met. S302: Draw the predicted QS curve: As various load levels are applied and the corresponding pile top settlement data is recorded, a predicted QS curve is plotted using graphing software, with the load value Q (kN) as the ordinate and the pile top settlement value S (mm) as the abscissa. This predicted QS curve intuitively reflects the trend of pile top settlement as load increases under theoretical loading conditions, based on the constructed model and parameters.

[0022] S4: Draw the actual QS curve: S401: Field test setup. At the pile construction site, a high-precision displacement sensor was installed at the pile top to ensure that the sensor could accurately measure the vertical displacement of the pile top. The displacement sensor was connected to a data acquisition device via a data cable. An appropriate data acquisition frequency was set to record the changes in pile top settlement in real time. Hydraulic jacks were used to apply staged loading to the pile top, and the loading equipment was securely connected to the reaction device. The reaction device could utilize the reaction force provided by nearby completed foundation piles or ground anchors to ensure a stable and safe loading process. During the field test, the actual loading levels were consistent with those used in the numerical simulation.

[0023] S402: Field test operation and data collection: Start the loading test and apply vertical load to the pile top step by step according to the predetermined loading levels and loading amounts. After applying each level of load, wait for a certain period of time to allow the deformation of the pile body and the soil to fully develop until the settlement rate stabilizes within a certain allowable range. At this time, the pile top settlement value measured by the displacement sensor is recorded by the data acquisition instrument. Repeat this process until all loading levels are completed.

[0024] S403: Draw the actual QS curve: The load values and corresponding pile top settlement data collected from the field tests were collated and the actual QS curve was plotted, again with load Q as the ordinate and pile top settlement S as the abscissa. The actual QS curve reflects the bearing-deformation characteristics of foundation piles under actual engineering conditions.

[0025] S5: Generate the corrected pile side resistance and pile end resistance: S501: Calculate soil influence coefficient: Compare the predicted QS curve to the actual QS curve and calculate the deviation between the two. This deviation can be measured in a variety of ways, such as averaging the difference in pile top settlement at the same load point between the two curves, or quantifying the degree of deviation using a goodness-of-fit index.

[0026] Assuming that the average value of the settlement difference between the two curves under various load levels is ΔS, the soil influence coefficient is defined as for: ; in It is the average value of the pile top settlement in a certain load segment in the actual Q-S curve. This coefficient reflects the degree of influence on the bearing capacity of the foundation pile due to the difference between the actual soil conditions and the model assumptions. S502: Correct the pile side resistance and pile end resistance according to the soil influence coefficient: Calculation of corrected pile side wear resistance and corrected pile tip resistance : , ,in, is the side wear resistance of the original pile, is the original pile end resistance S503: Calculate the vertical bearing capacity using the corrected pile side resistance and pile end resistance: Vertical bearing capacity of pile foundation The calculation formula is: The vertical bearing capacity is calculated using the corrected pile side resistance and pile end resistance, and the calculated pile bearing capacity is corrected according to the influence coefficient of the environment during the actual test process, thereby reducing the systematic deviation between the theoretically calculated ultimate load and the actual failure load, and improving the accuracy of the pile bearing capacity calculation.

[0027] S61: Build and train LSTM network: S611: Building an LSTM network: Select a suitable deep learning framework, such as TensorFlow or PyTorch, and build a neural network model that includes an input layer, one or more LSTM hidden layers, and an output layer. The number of input layer nodes is determined according to the dimension of the input features. For example, in this embodiment, the input features include multiple sets of actual test data with 10 point elements (20 dimensions, 10 for load and settlement), 5 standardized soil parameters, and 10 settlement acceleration values. Then the number of input layer nodes is 35. The number of hidden layer LSTM units can be determined through experimental debugging and is initially set to 64 units. The front and back layers are connected in a fully connected manner, and a suitable activation function is set, such as the tanh function for updating the internal state of the LSTM unit. The output layer uses a linear activation function, and the number of output nodes is 1, corresponding to the settlement prediction value of the future first-level load. S612: Configure training parameters: Select a suitable optimization algorithm, such as the Adam optimization algorithm, set the initial learning rate to 0.001, and use a learning rate decay strategy of 0.1 every 10 epochs (training rounds).

[0028] At the same time, the loss function is defined. Since this is a regression prediction problem, the mean square error (MSE) is selected as the loss function to measure the error between the predicted value and the true value: ; in, is the predicted value, is the true value, N is the number of samples; S613: Start training: Input the training set data into the constructed LSTM network and perform multiple rounds of iterative training according to the set training parameters; In each round of training, the model predicts the settlement value for the next level of load based on the input sample characteristics, calculates the loss between the predicted value and the true label, and then adjusts the network weight parameters through backpropagation through an optimization algorithm to minimize the loss function. After hundreds of rounds of training, until the loss on the test set no longer decreases significantly, the model reaches a state of convergence. At this point, the model has learned the complex nonlinear relationship between the actual QS curve, soil parameters, and changes in pile top settlement. S63: Forecast and warning departure: S631: Prediction: After completing training and verifying the model's effectiveness, the actual test data collected in real time is processed according to the preprocessing method described above and input into the trained LSTM network for prediction. For example, when loading on site to the 8000kN level (corresponding to the first 80 points of the actual QS curve), the 10 most recent points and their corresponding soil parameters and settlement acceleration are selected and input into the LSTM network. The model then outputs the predicted pile top settlement value for the next level of load (i.e., the 9000kN level). As loading continues, the input samples are continuously updated to predict the pile top settlement trend under subsequent levels of load. S632: Set the warning threshold. This threshold is determined based on project experience and previous analysis of pile bearing characteristics. For example, based on empirical data from similar projects and the design requirements of this project, a warning is triggered when the predicted increase in pile top settlement exceeds 10mm under n future load levels (assuming n=3).

[0029] During the prediction process, the settlement prediction values of the future n levels of load output by the model are monitored in real time, and the settlement increment between two adjacent levels of load is calculated. Once the settlement increment is found to exceed the warning threshold, a warning signal is immediately sent to on-site construction personnel and engineering technicians through sound and light alarms, SMS notifications, etc., to remind them that there may be a bearing risk in the foundation piles and that loading needs to be suspended or further inspection measures need to be taken to ensure project safety.

[0030] By leveraging the time series prediction capabilities of the LSTM network, combined with field test data and pre-calculated soil parameters, we were able to accurately predict the settlement trend of pile tops with enlarged heads during subsequent loading. This also enabled us to trigger early warnings in a timely and effective manner, providing strong support for pile construction quality control and engineering safety.

[0031] Among them, in S501, the offset between the predicted QS curve and the actual QS curve can be obtained by comparing the predicted settlement corresponding to the predicted QS curve and the actual QS curve step by step. and actual settlement , calculate the root mean square error; , n is the total number of load stages; Calculate the soil influence coefficient: , is the settlement threshold allowed by the test specification.

[0032] In order to, in one embodiment, in step S1, use the Coulomb friction model to set the contact surface between the soil layer and the pile body, the friction coefficient , is the pile-soil interface friction angle, which includes the following steps: a: Experimental preparation stage: Pile specimens: Use concrete of the same grade as the actual test project to make cylindrical specimens. Before the initial setting of the concrete, quartz sand with a particle size of 1-3mm is embedded in the pile surface to form an interface with a surface roughness similar to that of the cast-in-place pile on site. Soil sample: The original soil sample (such as silty clay) is taken from the main pile position drill hole and prepared by the ring knife method to ensure that its natural moisture content and dry density are consistent with the in-situ soil layer actually tested on site; Use an improved direct shear apparatus equipped with high-precision force sensors and displacement sensors; A custom annular shear box is designed with soil samples placed in the inner layer and pile samples placed in the outer layer to simulate the annular contact between the pile and soil. b: Experimental stage: Install the sample: fix the pile sample vertically on the base of the shear box, and put the soil sample on the outside of the pile to ensure that the contact surface is tightly fitted; Gradual application of normal stress: according to the actual soil distribution of the project, set multiple groups of normal stress , respectively set to 50kPa, 100kPa, 150kPa (corresponding to the overburden pressure at the pile depth of about 3m, 6m, and 9m; Apply horizontal shear displacement at a rate of 0.02 mm / min, and record the shear force (F) and displacement ( ) data until the shear displacement reaches 10 mm or the shear force reaches a peak and then steadily decreases; The interface shear strength is calculated based on the test data. , A is the contact area between the pile and the soil, draw , fitting the Coulomb formula by linear regression: , The slope of , the intercept is the soil cohesion c.

[0033] In constructing the three-dimensional pile-soil numerical model, the pile-soil interface is defined using CONTACT PAIR, and the friction properties are set to the Coulomb friction model. Through customized interface direct shear tests, the Coulomb friction parameters of the pile-soil interface are quantitatively measured, and the engineering applicability of the parameters is ensured through model verification. In the numerical simulation, the contact surface parameters consistent with the field test are used, showing In one embodiment, the LSTM network is used to predict the pile top settlement trend. When m is the predicted load level n in the future, the early warning is triggered. Specifically, the following steps are included: Collect and organize multiple sets of actual test data obtained from the field test in step S4, each set of actual test data contains a corresponding load value Q and pile top settlement value S, and make it into a two-dimensional data matrix, where the first column stores the load value and the second column stores the corresponding settlement value; Combined with the soil parameters measured and calculated in step S1, such as the internal friction angle, cohesion, compression modulus, etc. of different soil layers, these parameters are standardized. The purpose of standardization is to make different parameters have a unified dimension and numerical range, which is convenient for the subsequent training and calculation of the neural network model. For example, using the Z-score standardization method, for a certain soil parameter X, calculate its mean and standard deviation , convert the original value X into a standardized value ; Calculate the pile top settlement acceleration. The pile top settlement acceleration is calculated by numerically differencing the actual QS curve. Specifically, the central difference method is used to calculate the pile top settlement acceleration for the discrete pile top settlement sequence. , m is the total number of load levels applied to predict the QS curve, At some point Sedimentation velocity The calculation formula is: ; in Q is the load step length. Similarly, the calculated settlement acceleration series is normalized; The multiple sets of actual test data, standardized soil parameters, and settlement accelerations that have undergone the above processing are combined in chronological order to form the input sample dataset for LSTM network training. For example, every 10 consecutive point elements and their corresponding soil parameters and settlement accelerations are used as an input sample. The sample label is the settlement value corresponding to the subsequent 11th point element (i.e., the predicted settlement value for the next level of load). The training set is constructed in this way, and the test set uses the point elements of the predicted QS curve. The accuracy of the predicted QS curve is significantly improved, providing a reliable basis for subsequent bearing capacity correction.

[0034] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as above in terms of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can, without departing from the scope of the technical solution of the present invention, make some changes or modifications to equivalent embodiments using the technical contents disclosed above. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A method for testing the vertical compressive bearing capacity of piles with enlarged heads, characterized by: The following steps are involved: S1: Obtaining measurement data, the measurement data including soil layer parameters, pile structure parameters, and contact surface parameters, and constructing a three-dimensional pile-soil layer numerical model based on the measurement data; S2: Establish the load loading model and complete the pile bottom displacement reset; S3: Use the theoretically set load value and loading level to apply load step by step, obtain the pile top settlement corresponding to each load level, and draw the predicted QS curve; S4: Obtain the pile top settlement corresponding to each load level in the field test and draw the actual QS curve; S5: Calculating a soil influence coefficient based on the deviation between the predicted QS curve and the actual QS curve, correcting the pile side resistance and pile end resistance based on the soil influence coefficient, and calculating the vertical bearing capacity using the corrected pile side resistance and pile end resistance.

2. A method for testing the vertical compressive bearing capacity of piles with enlarged heads according to claim 1, characterized in that: In step S1, the soil layer parameters include the density, Poisson's ratio, cohesion, internal friction angle, elastic modulus and stiffness of the rock layer and soil layer at the pile implantation depth of the test point; The pile body structural parameters include the length and cross-sectional area of the pile body with a constant cross section, and the height and shape parameters of the pile body with a variable cross section.

3. A method for testing the vertical compressive bearing capacity of piles with enlarged heads according to claim 1, characterized in that: Step S5 includes the following sub-steps: S51: Compare the predicted settlement corresponding to the predicted QS curve and the actual QS curve step by step and actual settlement , calculate the root mean square error; , n is the total number of load stages; S52: Calculate soil influence coefficient: , is the settlement threshold allowed by the test specification; S53: Calculate and correct pile side wear resistance and corrected pile tip resistance : , ,in, is the side wear resistance of the original pile, is the original pile end resistance; S54: Get the corrected vertical bearing capacity The calculation formula is: .

4. A method for testing the vertical compressive bearing capacity of piles with enlarged heads according to claim 1, characterized in that: Step S5 is followed by S6: using LSTM network to predict the pile top settlement trend, and When , an early warning is triggered, m is the predicted future n-th level load, where the training set of the LSTM network is the point element of the actual QS curve, and the validation set is the point element of the predicted QS curve.

5. A method for testing the vertical compressive bearing capacity of piles with enlarged heads according to claim 4, characterized in that: The input features of the LSTM network include point elements of the actual QS curve, calculated soil parameters and calculated pile top settlement acceleration, and the output is the settlement prediction value of the future n-level load.

6. A method for testing the vertical compressive bearing capacity of piles with enlarged heads according to claim 5, characterized in that: The training of the LSTM network includes: Data processing: the load value Q and pile top settlement value in the actual test are standardized using the Z-score standardization method; The actual QS curve is numerically differentiated to calculate the pile top settlement acceleration; The standardized data are combined in the order of the time series to form the input sample data set for LSTM network training; Build and train the LSTM network and define the mean square error function as the loss function; The training set data is input into the constructed LSTM network, and multiple rounds of iterative training are performed according to the set training parameters. The training parameters are the point set of the actual QS curve. The weight parameters of the network are then adjusted through back propagation of the optimization algorithm to minimize the loss function. The actual test data collected in real time is processed according to the above preprocessing method and input into the trained LSTM network for prediction.

7. A method for testing the vertical compressive bearing capacity of piles with enlarged heads according to claim 1, characterized in that: In step S1, the contact surface between the soil layer and the pile body is set using the Coulomb friction model, and the friction coefficient , is the friction angle between the pile and soil, which is determined through interface tests.

8. A method for testing the vertical compressive bearing capacity of piles with enlarged heads according to claim 7, characterized in that: The determination of the Coulomb model includes: Install the sample: fix the pile sample vertically on the base of the shear box, and put the soil sample on the outside of the pile to ensure that the contact surface is tightly fitted; Gradual application of normal stress: according to the actual soil distribution of the project, set multiple groups of normal stress ; Calculate the interface shear strength using the formula: , A is the contact area between the pile and the soil, draw , fitting the Coulomb formula by linear regression: , The slope of , the intercept is the soil cohesion c.

Images