Pile load transfer simulation method and system combined with machine learning

Abstract

The present application provides a kind of pile foundation load transfer simulation method and system combined with machine learning, first, the interaction scene model including pile foundation construction element, soil level element and external load element is built, and load transfer process fragment set under different load conditions is generated, then the pre-trained load transfer dynamic evolution model is called to carry out inversion learning to fragment set, generate load transfer path and pile-soil interface state evolution characteristic sequence, then according to load transfer path and pile-soil interface state evolution characteristic sequence, dynamic deduction is carried out, the dynamic characteristics of pile foundation load distribution and pile displacement are obtained, finally, based on these dynamic characteristics, simulation analysis report containing load transfer dynamic process curve is generated, thereby the accuracy and reliability of pile foundation load transfer simulation are improved.

Description

Technical Field

[0001] This invention relates to the field of machine learning technology, and more specifically, to a method and system for simulating pile foundation load transfer by incorporating machine learning. Background Technology

[0002] In the field of civil engineering, pile foundations, as the basic supporting structure of buildings, are crucial to the stability and safety of buildings due to their load transfer characteristics. Traditional methods for analyzing the load transfer of pile foundations mainly rely on theoretical formula derivation and field tests. Theoretical formulas are often based on some simplifying assumptions and are difficult to accurately reflect the complex interaction between the pile foundation and the soil; while field tests suffer from problems such as high cost, long cycle, and limitations imposed by site conditions, and are difficult to comprehensively obtain dynamic information on the load transfer of pile foundations under different load conditions.

[0003] With the development of computer technology, numerical simulation methods have been gradually applied to pile foundation load transfer analysis. However, most existing numerical simulation methods are based on fixed physical models and parameters, lacking the ability to adaptively learn and dynamically extrapolate complex and ever-changing situations in actual engineering, and thus cannot accurately simulate the dynamic evolution process of pile foundation load transfer. Summary of the Invention

[0004] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide a pile foundation load transfer simulation method incorporating machine learning, the method comprising:

[0005] Construct an interaction scenario model between the pile foundation and the soil, wherein the interaction scenario model includes pile foundation structural elements, soil layer elements, and external load elements;

[0006] Based on the interaction scenario model, a set of load transfer process segments is generated, which contains continuous process records of the interaction between the pile foundation and the soil under different load conditions.

[0007] The pre-trained load transfer dynamic evolution model is invoked to perform inversion learning on the set of load transfer process segments, generating load transfer path evolution feature sequences and pile-soil interface state evolution feature sequences.

[0008] Based on the load transfer path evolution characteristic sequence and the pile-soil interface state evolution characteristic sequence, dynamic deduction of load transfer is performed to obtain the dynamic characteristics of pile foundation load distribution and pile displacement evolution characteristics.

[0009] Based on the dynamic characteristics of the pile foundation load distribution and the evolution characteristics of the pile body displacement, a pile foundation load transfer simulation analysis report containing load transfer dynamic process curves is generated.

[0010] In another aspect, embodiments of the present invention also provide a pile foundation load transfer simulation system incorporating machine learning, including a processor and a machine-readable storage medium connected to the processor. The machine-readable storage medium is used to store programs, instructions, or code, and the processor is used to execute the programs, instructions, or code in the machine-readable storage medium to implement the above-described method.

[0011] Based on the above, a detailed interaction scenario model between the pile foundation and the soil is constructed, comprehensively covering key elements such as pile foundation structure, soil layers, and external loads. Based on this interaction scenario model, a set of load transfer process segments is generated, and a pre-trained dynamic evolution model of load transfer is used for inverse learning. This allows for in-depth exploration of the evolutionary characteristics of the load transfer path and the pile-soil interface state. Through dynamic simulation, the dynamic characteristics of pile foundation load distribution and pile displacement are obtained, and a simulation analysis report containing dynamic process curves is generated. This achieves accurate simulation and comprehensive analysis of the pile foundation load transfer process, thereby improving the accuracy and reliability of pile foundation load transfer analysis. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of the execution flow of the pile foundation load transfer simulation method combined with machine learning provided in an embodiment of the present invention.

[0013] Figure 2 This is a schematic diagram of exemplary hardware and software components of the pile foundation load transfer simulation system combined with machine learning provided in an embodiment of the present invention. Detailed Implementation

[0014] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating a pile foundation load transfer simulation method incorporating machine learning, as provided in one embodiment of the present invention. The following is a detailed description of this pile foundation load transfer simulation method incorporating machine learning.

[0015] Step S110: Construct an interaction scenario model between the pile foundation and the soil, wherein the interaction scenario model includes pile foundation structural elements, soil layer elements, and external load elements.

[0016] In this embodiment, the interaction between a precast concrete pile and the surrounding multi-layered soil in a construction project is used as the application scenario to construct an interaction scenario model. First, the structural elements of the pile foundation are defined. The pile foundation is cylindrical and consists of three physical components: the pile body, the pile tip, and the pile top. The material used for the pile body has a specific elastic modulus and Poisson's ratio. The pile tip is a conical structure, and the pile top is equipped with a connecting structure to bear external loads.

[0017] Next, the soil layer elements are determined. In this scenario, the soil is divided into three layers from the surface downwards: the first layer, the second layer, and the third layer. Each layer has different physical and mechanical properties such as unit weight, internal friction angle, and cohesion. The thickness of each soil layer is also different, and there are obvious interfaces between the soil layers.

[0018] Next, consider the external load elements. The external load is mainly a vertical load, which can be applied in two ways: static load and variable load. Static load is a constant load that acts continuously, while variable load exhibits a certain pattern of change over time, such as periodic or linear changes.

[0019] By integrating the above-mentioned pile foundation structural elements, soil layer elements, and external load elements, a complete interaction scenario model is constructed. This interaction scenario model can reflect the interaction of the pile foundation under external loads in a specific soil environment.

[0020] Step S111: Determine the spatial boundary of the interaction between the pile foundation and the soil, wherein the spatial boundary includes the spatial area where the pile foundation is located and the spatial area affected by the soil.

[0021] When constructing an interaction scenario model, the spatial boundaries must first be determined. For the spatial region where the pile foundation is located, taking the central axis of the pile foundation as a reference, and considering the diameter and length of the pile foundation, its coordinate range in three-dimensional space is determined. This coordinate range in three-dimensional space should completely encompass the entire pile foundation structure.

[0022] For the spatial area affected by the soil, a sufficiently large spatial range is determined based on the length of the pile foundation and the properties of the surrounding soil. This range allows for the full consideration of the stress and deformation of the soil within it, while the influence of soil outside this range on the pile foundation can be ignored. For example, in the transverse direction, the range extends a certain distance to both sides of the central axis of the pile foundation; in the longitudinal direction, it extends downwards from the ground surface to a certain depth below the pile tip. Thus, by defining the boundaries of this spatial range, a clear area is delineated for subsequent model construction and simulation calculations.

[0023] Step S112: Divide the space into pile foundation structural units and soil layer units. The pile foundation structural units correspond to the physical components of the pile foundation, and the soil layer units correspond to the physical components of different soil layers.

[0024] Within the defined spatial boundaries, the pile foundation is divided into structural units. The pile body is divided along its length into multiple cylindrical pile foundation structural units, each with the same length and diameter, and whose material properties are consistent with the overall pile body. The pile tip and pile top are also divided into independent pile foundation structural units to accurately reflect their specific structural and stress conditions.

[0025] For soil hierarchical units, the soil influence area is divided into soil hierarchical units corresponding to each soil layer according to the interfaces of the soil layers. Each soil hierarchical unit has certain dimensions in both the transverse and longitudinal directions, and soil hierarchical units within the same soil layer have the same physical and mechanical properties. For example, the dimensions of the soil hierarchical unit corresponding to the first soil layer are determined according to the distribution range of that soil layer, and the division of soil hierarchical units for the second and third soil layers is similar. Through the above division, the pile foundation and soil can be analyzed and calculated in the form of units.

[0026] Step S113: Set the interaction rules between the pile foundation structural unit and the soil layer unit. The interaction rules are used to describe the specific ways in which the pile foundation interacts with the surrounding soil when it is under load.

[0027] When setting the interaction rules, the contact relationship between the pile foundation structural unit and the adjacent soil layer unit is considered. When the pile foundation is subjected to load, the pile structural unit will exert pressure on the surrounding soil layer unit, and at the same time, the soil layer unit will generate a reaction force on the pile structural unit. The above interaction is manifested in the form of friction and normal force.

[0028] For the relationship between the pile foundation structural unit and the soil layer unit on the pile side, the main consideration is the effect of friction. The magnitude of the friction is related to the contact area between the pile structural unit and the soil layer unit, the internal friction angle of the soil, and the normal pressure on the contact surface. For the relationship between the pile tip structural unit and the corresponding soil layer unit, the main consideration is the effect of normal force. The magnitude of the normal force is related to the load transmitted by the pile tip and the bearing capacity of the soil.

[0029] Meanwhile, it is set that when the pile foundation structural unit is displaced, the soil layer unit will produce corresponding deformation, which will in turn affect the stress state of the pile foundation structural unit, forming a dynamic interaction relationship.

[0030] Step S114: Configure the initial state parameters of the pile foundation structural unit and the initial state parameters of the soil layer unit. The initial state parameters include the initial stress and initial displacement conditions.

[0031] When configuring the initial state parameters of the pile foundation structural unit, the initial stress condition for each pile foundation structural unit is zero, meaning that there is no stress inside each unit when no external load is applied. The initial displacement condition is also zero; each pile foundation structural unit is in its initial position without any displacement.

[0032] For the soil layer unit, the initial stress condition considers the self-weight stress of the soil. Soil layer units at different depths have different self-weight stresses; the greater the depth, the greater the self-weight stress. Its magnitude is related to the unit weight and thickness of the soil above the unit. The initial displacement condition is also zero; the soil layer unit is in a stable position in the initial state and no displacement occurs.

[0033] These initial state parameters are input into the interaction scenario model as initial conditions for the model simulation.

[0034] Step S115: Verify the validity of the interaction scenario model to ensure that the spatial boundary, the pile foundation structural unit, the soil layer unit, the interaction mode rules, and the initial state parameters are mutually compatible.

[0035] When validating the interaction scenario model, first check whether the spatial boundary can completely include the soil area that the pile foundation may affect when bearing the maximum load. If the boundary is found to be too small and cannot cover the entire affected area, the boundary needs to be adjusted.

[0036] Next, verify whether the division of the pile foundation structural unit and the soil layer unit is reasonable, and ensure that the size of the unit can accurately reflect the stress and deformation characteristics of the structure and soil. If the unit division is too large, resulting in insufficient accuracy, or too small, increasing the amount of calculation, then make the corresponding adjustments.

[0037] Then, check whether the interaction rules match the actual physical and mechanical properties of the pile foundation and soil. By comparing the interaction rules with those in similar existing engineering cases, verify the rationality of the rules.

[0038] Finally, verify whether the initial state parameters are configured correctly, and ensure that the initial stress and initial displacement settings are consistent with the actual situation, such as whether the initial self-weight stress calculation of the soil layer unit is accurate.

[0039] Through the above verification steps, it is ensured that the various components of the model are compatible with each other and can accurately simulate the interaction between the pile foundation and the soil.

[0040] Step S120: Generate a load transfer process segment set based on the interaction scenario model. The load transfer process segment set contains continuous process records of the interaction between the pile foundation and the soil under different load conditions.

[0041] After constructing and validating the interaction scenario model, a set of load transfer process segments is generated based on the model. By setting different load conditions, the model is run to simulate and record the continuous process of interaction between the pile foundation and the soil. These process records are then processed to form a set of process segments to comprehensively reflect the load transfer characteristics under different load conditions.

[0042] Step S121: Set multiple load conditions in the interaction scenario model. The load conditions include different load application locations, different load application methods, and different load variation forms.

[0043] Multiple load conditions are set in the interaction scenario model. The load is mainly applied at the top of the pile, but the case of applying lateral loads at different heights of the pile is also considered.

[0044] The load application methods include concentrated loads and uniformly distributed loads. Concentrated loads act on a specific point at the top of the pile, while uniformly distributed loads are evenly distributed over a certain area at the top of the pile.

[0045] There are many forms of load variation, such as static load, which remains constant after the load is applied; linearly increasing load, which gradually increases from the initial value to a certain value and then remains stable; and periodically changing load, which changes according to a certain period and amplitude.

[0046] By combining the different load application locations, application methods, and variations mentioned above, various load action conditions are formed for simulating the interaction scenario model.

[0047] Step S122: For each of the load conditions, run the interaction scenario model to simulate and record all process information of the pile foundation from the start of stress to reaching a stable state. The all process information includes the load transfer of each part of the pile foundation and the stress response of each unit of the soil.

[0048] For each load condition, an interaction scenario model is run for simulation. During the simulation, starting from the moment the load is applied, the load transfer situation of each part of the pile foundation is recorded at regular intervals, including the magnitude and direction of the load borne by each structural unit of the pile top, pile body, and pile tip, as well as the load transfer path between units.

[0049] At the same time, the stress response of each soil unit is recorded, including the magnitude and direction of the stress on each soil layer unit and the deformation of the unit.

[0050] Record continuously until the pile foundation and soil reach a stable state, that is, the load and displacement of each part of the pile foundation no longer change with time, and the stress and deformation of each unit of the soil also tend to stabilize. At this point, stop recording and obtain all process information under the load conditions.

[0051] Step S123: The entire process information is segmented according to time sequence, and the entire process information is divided into multiple continuous process segments, each of which corresponds to a specific stage in the load transfer process.

[0052] When processing all process information in segments according to time sequence, the segmentation points are determined based on the characteristic changes during load transfer. For example, in the stage of linear load increase, when the load increases to a certain proportion, it can be considered a segmentation point; in the stage where the load remains constant, when the deformation rate of the pile foundation and soil changes significantly, it can also be considered a segmentation point.

[0053] The entire process information from the beginning to the steady state is divided into multiple continuous process segments according to these segmentation points. Each process segment corresponds to a specific stage, such as the initial stage of load increase, the middle stage of load increase, the initial stage of load stabilization, and the late stage of load stabilization.

[0054] Each process segment contains records of pile load transfer and soil stress response at all points in time within that stage.

[0055] Step S124: Identify key change nodes in each process segment, including nodes where the load transfer direction changes and nodes where the pile-soil interface interaction intensity changes.

[0056] When identifying key change nodes in each process segment, for nodes where the load transfer direction changes, by analyzing the load transfer direction records of each structural unit of the pile foundation in the process segment, when the load transfer direction in a certain unit changes significantly compared to the previous moment, for example, from vertical transfer to both vertical and horizontal transfer, that moment is a key change node.

[0057] For nodes where the interaction strength between the pile and the soil interface changes, monitor the magnitude of the frictional force and normal force between the pile structural unit and the soil layer unit. When the magnitude of these forces changes abruptly at a certain moment, that is, when the interaction strength changes, that moment is another key change node.

[0058] These key change nodes are marked in the corresponding process segments for subsequent analysis and processing.

[0059] Step S125: The process segments containing the key change nodes are classified and organized according to the load application conditions to form a load transfer process segment set with temporal sequence correlation. Each process segment in the load transfer process segment set contains complete load transfer stage characteristics.

[0060] The process segments containing key change nodes are classified according to the load conditions. For example, all process segments under static load conditions are grouped into one category, and process segments under linearly increasing load conditions are grouped into another category, and so on.

[0061] In each category, the process segments are arranged in chronological order so that the end time of the previous process segment is connected to the start time of the next process segment, forming a sequence with temporal correlation.

[0062] The resulting set of load transfer process segments contains complete load transfer characteristics for each corresponding stage, including information such as load magnitude, transfer direction, and pile-soil interaction intensity, and is arranged in an orderly manner according to load conditions and time sequence.

[0063] Step S130: Call the pre-trained load transfer dynamic evolution model to perform inversion learning processing on the set of load transfer process segments, and generate load transfer path evolution feature sequence and pile-soil interface state evolution feature sequence.

[0064] The pre-trained load transfer dynamic evolution model is invoked, and the generated set of load transfer process segments is input into the model. The model analyzes the laws of load transfer and the changes in the pile-soil interface state through inverse learning of the set of process segments, and then generates the load transfer path evolution feature sequence and the pile-soil interface state evolution feature sequence to comprehensively reflect the dynamic process of load transfer.

[0065] Step S131: Input the load transfer process segment set into the process coding component of the load transfer dynamic evolution model, encode and convert the time series data in the load transfer process segment set to generate a process coding data sequence, which is used to represent the time series characteristics of the load transfer process.

[0066] The set of load transfer process segments is input into the process coding component of the load transfer dynamic evolution model. The process coding component first parses the time series data in each process segment and extracts characteristic parameters such as load magnitude, transfer direction, and pile-soil interaction intensity.

[0067] Then, these feature parameters are encoded and converted into numerical forms suitable for model processing. For example, the load transfer direction is converted into corresponding angle values, and the interaction intensity is converted into corresponding dimensionless values.

[0068] The encoded feature parameters are arranged in chronological order to form the encoded data of each process segment. Then, the encoded data of all process segments are connected in chronological order to generate the process encoded data sequence of the entire load transfer process. This process encoded data sequence can completely represent the time series characteristics of the load transfer process.

[0069] Step S132: The path feature of the process encoded data sequence is captured by the path tracking component of the load transfer dynamic evolution model, and the changes in the load transfer path inside the pile foundation and between the pile foundation and the soil are analyzed to generate the initial features of the load transfer path.

[0070] The process-coded data sequence is processed using a path tracing component of the load transfer dynamic evolution model. The path tracing component analyzes the process-coded data sequence segment by segment to identify the path of load transfer within the pile foundation from the pile top to the pile body and pile tip, as well as the path of load transfer from the pile body to the surrounding soil.

[0071] During the analysis, pay attention to the changes in the direction and range of the load transfer path. For example, when the load is transferred through the pile, does it maintain a straight line or does it deviate to one side? When the load is transferred to the soil, does the range of transfer gradually expand or remain within a certain range?

[0072] Based on these analysis results, information that can characterize the path features is extracted, such as the main direction of the path, the covered pile foundation structural units and the range of soil layer units, etc., and combined to form the initial features of the load transfer path.

[0073] Step S1321: Set a path identification window in the path tracing component. The path identification window is used to extract local time series segments in the process encoded data sequence.

[0074] In the path tracing component, a path identification window is set, which has a certain time length. For example, the time length of the window can cover several consecutive time points in the process-encoded data sequence, so that the window can capture local time series segments with a certain time span.

[0075] The window size can be set according to the time resolution of the process-encoded data sequence to ensure that the extracted local time series segments can reflect the basic characteristics of the load transfer path within that time period.

[0076] Step S1322: Traverse the process-encoded data sequence through the path identification window, determine the path direction for each local time series segment, and determine the transmission direction and coverage of the load within the local time series segment.

[0077] The process-encoded data sequence is traversed through a path recognition window. Starting from the beginning time point of the sequence, the window moves forward one time point at a time until the entire sequence has been traversed.

[0078] For each captured local time series segment, the path tracing component analyzes the relevant data on load transfer within it to determine the direction of load transfer during that time period, such as whether it is transferred vertically downwards, horizontally, or diagonally.

[0079] At the same time, the load transfer coverage area is determined, that is, the number and location range of pile foundation structural units and soil layer units involved in the load transfer, such as which structural units of the pile body and which surrounding soil layer units are covered.

[0080] Step S1323: Compare and analyze the transmission direction and coverage of adjacent local time series segments, analyze the continuation and turning points of the load transmission path, and generate path change identifiers.

[0081] Compare the transmission direction and coverage of two adjacent local time series segments. If the transmission direction of the latter segment is basically the same as that of the former segment, and the transmission coverage is also basically the same or shows continuous expansion or contraction, it indicates that the load transmission path is continuous.

[0082] If the direction of load transfer in the next segment differs significantly from that of the previous segment, or if the coverage area of ​​the transfer changes discontinuously, it indicates that the load transfer path has changed direction.

[0083] Based on the comparative analysis results, path change identifiers are generated. Continuation scenarios are marked with a specific identifier, while turning scenarios are marked with another specific identifier to distinguish different path change situations.

[0084] Step S1324: The process-encoded data sequence is segmented according to the path change identifier to obtain multiple path segments with stable transmission paths.

[0085] The process coded data sequence is segmented based on the path change identifier. When a path change identifier is encountered in a turning situation, the time point where the identifier is located is taken as the segmentation point.

[0086] In this way, the process-coded data sequence is divided into multiple continuous parts, each of which contains several continuous local time series segments, and the load transfer path within each part is in a continuous state, that is, it has a stable transfer direction and transfer coverage. These parts are the path segments with stable transfer paths.

[0087] Step S1325: Extract the transfer direction features, transfer coverage features and duration features of each path segment and combine them to form the initial features of the load transfer path. The initial features of the load transfer path are used to represent the basic features of the load transfer path at different stages.

[0088] For each path segment, its load transfer direction characteristics are extracted, that is, the main direction of load transfer within the segment. This can be determined by comprehensively considering the load transfer directions of each local time series segment within the segment.

[0089] Extract the load transfer coverage characteristics, that is, the overall range of pile foundation structural units and soil layer units involved in the load transfer within this segment.

[0090] Extract duration features, which is the length of time the path segment takes from start to finish.

[0091] These three features are combined to form the initial features of the load transfer path for that path segment. The initial features of multiple path segments are arranged in chronological order to jointly constitute the initial features of the load transfer path for the entire process.

[0092] Step S133: Use the state association component of the load transfer dynamic evolution model to perform association analysis on the key change nodes in the process coded data sequence, determine the association relationship between the pile-soil interface interaction intensity and the load transfer path, and generate an association relationship feature description.

[0093] By utilizing the state association component of the load transfer dynamic evolution model, relevant data of all key change nodes are extracted from the process-encoded data sequence. These data include the pile-soil interface interaction intensity and load transfer path characteristics at the key change nodes.

[0094] The state correlation component analyzes this data to study how changes in the pile-soil interface interaction intensity affect changes in the load transfer path at key change nodes, and how changes in the load transfer path, in turn, affect changes in the pile-soil interface interaction intensity. For example, when the pile-soil interface interaction intensity increases, it analyzes whether the load transfer path becomes more concentrated or whether the transfer direction deviates; when the load transfer path extends to new soil layer units, it analyzes whether the pile-soil interface interaction intensity changes accordingly. Through this bidirectional analysis, the correlation pattern between the two is clarified, thereby generating a correlation feature description that can quantify the above-mentioned correlation.

[0095] Step S1331: Extract state description data of all key change nodes from the process-encoded data sequence. The state description data includes pile-soil interface interaction strength parameters and load transfer path parameters.

[0096] When extracting state description data for all key change nodes from the process-coded data sequence, the position of each key change node in the sequence is first located. For each key change node, its corresponding pile-soil interface interaction strength parameters are extracted. These parameters include the magnitude of the pile side friction force and the magnitude of the pile tip normal force, which reflect the interaction strength of the pile-soil interface at that node.

[0097] At the same time, load transfer path parameters at this key change node are extracted. These load transfer path parameters include the main direction angle of load transfer, the range information of the pile foundation structural units and soil layer units covered by the transfer, etc. These parameters can describe the characteristics of the load transfer path at this node.

[0098] The extracted pile-soil interface interaction strength parameters and load transfer path parameters are combined to form state description data for each key change node.

[0099] Step S1332: Perform standardization transformation on the state description data, input the standardized state data into the correlation analysis model of the state association component, and calculate the mutual information value and correlation coefficient value between the pile-soil interface interaction strength parameter and the load transfer path parameter.

[0100] The condition description data undergoes standardization transformation to convert the pile-soil interface interaction strength parameters and load transfer path parameters to the same numerical range, eliminating the influence of differences in dimensions and numerical ranges between different parameters. For example, parameter values ​​are converted into values ​​within a specific range according to set rules.

[0101] The standardized state data is input into the correlation analysis model of the state correlation component. This model performs pairwise analysis on the pile-soil interface interaction strength parameters and load transfer path parameters. The mutual information value between the two parameters is calculated; a larger value indicates a stronger dependency. The correlation coefficient value is also calculated; this value measures the degree of linear correlation between the two parameters. The coefficient ranges within a specific interval, with positive values ​​indicating positive correlation and negative values ​​indicating negative correlation. A larger absolute value indicates a higher degree of correlation.

[0102] Step S1333: Determine the degree of correlation between each parameter based on the mutual information value and the correlation coefficient value, and filter out parameter combinations whose degree of correlation exceeds a preset threshold.

[0103] The degree of correlation between parameters is determined based on the calculated mutual information value and correlation coefficient value. The mutual information value and correlation coefficient value are combined for a comprehensive judgment; for example, when both the mutual information value and the absolute value of the correlation coefficient value are greater than a certain preset value, the corresponding parameter combination is considered to have a high degree of correlation.

[0104] A preset threshold is set, which is determined based on engineering experience and model training requirements. Parameter combinations with a correlation exceeding this preset threshold are selected; these parameter combinations reflect the important correlation between the pile-soil interface interaction intensity and the load transfer path.

[0105] Step S1334: Construct an association network structure based on the selected parameter combinations. In the association network structure, nodes represent parameters, edges represent the association relationships between parameters, and the weight of the edges represents the degree of association.

[0106] Based on the selected parameter combinations, a relational network structure is constructed. Each node in the network structure represents a parameter, including both the pile-soil interface interaction strength parameter and the load transfer path parameter.

[0107] For each selected parameter combination, an edge is created between the corresponding two nodes to indicate the correlation between the two parameters. The weight of the edge is determined based on the mutual information value and correlation coefficient value of the parameter combination; the stronger the correlation, the greater the weight of the edge.

[0108] The above method visually presents the relationships between parameters in the form of a network structure, forming a complete network structure of relationships.

[0109] Step S1335: Convert the relationship network structure into a relationship feature description containing node attributes and edge attributes. The relationship feature description is used to quantify the relationship between the pile-soil interface interaction intensity and the load transfer path.

[0110] When converting a network structure of relationships into a feature description of relationships, the attributes of each node in the network structure are first extracted, including the parameter name represented by the node and the physical meaning of the parameter.

[0111] Then, extract the attributes of each edge, including the names of the two nodes connected by the edge, the weight value of the edge, and the mutual information value and correlation coefficient value corresponding to the weight value.

[0112] These node and edge attributes are organized and recorded according to a defined format to form a relational feature description. This relational feature description can clearly quantify the relationship between the pile-soil interface interaction intensity and the load transfer path, including information such as which parameters are related and the degree of their relationship.

[0113] Step S134: Dynamically adjust the initial characteristics of the load transfer path based on the correlation feature description to obtain a load transfer path evolution feature sequence that can reflect the influence of pile-soil interaction.

[0114] The initial characteristics of the load transfer path are dynamically adjusted based on the correlation feature description. For example, when the correlation feature description indicates a strong correlation between a certain load transfer path parameter and the pile-soil interface interaction strength parameter, the path characteristics corresponding to that load transfer path parameter are adjusted according to the changes in the pile-soil interface interaction strength.

[0115] If the intensity of the pile-soil interface interaction increases, and the correlation shows that this will lead to a more concentrated load transfer path, then adjust the description of the transfer range in the initial characteristics of the load transfer path to reduce its range and make it more concentrated.

[0116] Through the above dynamic adjustments, the load transfer path characteristics can reflect the influence of pile-soil interaction. The adjusted path characteristics are arranged in chronological order to obtain the load transfer path evolution characteristic sequence.

[0117] Step S135: Extract the state characteristics of the pile-soil interface at different stages according to the load transfer path evolution characteristic sequence, and arrange them in chronological order to generate a pile-soil interface state evolution characteristic sequence. The pile-soil interface state evolution characteristic sequence includes the interface action type and action intensity characteristics at each stage.

[0118] Based on the load transfer path evolution sequence, the interaction between the load transfer path and the pile-soil interface is analyzed at different time stages. At each stage, the state characteristics of the pile-soil interface are extracted, including the type of interface interaction, such as whether it is primarily frictional force, normal force, or a combination of both.

[0119] Simultaneously, the interface interaction intensity characteristics are extracted, including the average magnitude and variation range of internal friction and normal forces during this stage. These state characteristics are arranged in chronological order to generate a pile-soil interface state evolution characteristic sequence, which fully presents the state changes of the pile-soil interface at different stages throughout the entire load transfer process.

[0120] Step S140: Perform dynamic simulation of load transfer based on the load transfer path evolution characteristic sequence and the pile-soil interface state evolution characteristic sequence to obtain the dynamic characteristics of pile foundation load distribution and pile displacement evolution characteristics.

[0121] Dynamic simulation of load transfer is performed using the load transfer path evolution characteristic sequence and the pile-soil interface state evolution characteristic sequence. During the simulation, the information provided by these sequences is combined to simulate the load transfer at different time stages, as well as the responses of the pile foundation and soil. This yields dynamic characteristics of pile foundation load distribution reflecting changes in pile load over time, and dynamic characteristics of pile displacement evolution reflecting changes in pile displacement over time.

[0122] Step S141: Input the load transfer path evolution characteristic sequence and the pile-soil interface state evolution characteristic sequence into the dynamic simulation module, initialize the simulation environment parameters, and set the simulation time interval and total simulation duration.

[0123] The load transfer path evolution characteristic sequence and the pile-soil interface state evolution characteristic sequence are input into the dynamic simulation module. The dynamic simulation module first initializes the simulation environment parameters, including setting the physical and mechanical property parameters of the pile foundation structural unit and soil layer unit, such as the elastic modulus of the pile foundation material and the unit weight of the soil. These parameters are consistent with the parameters in the interaction scenario model.

[0124] Set the simulation time interval, which is the time step for each simulation calculation. This interval should be small enough to ensure that subtle changes during load transfer are accurately captured. Set the total simulation duration, which should cover the entire process from the start of load application to the pile foundation and soil reaching a stable state.

[0125] Step S142: Within each simulation time interval, determine the load transfer direction and transfer ratio in the pile foundation and soil based on the load transfer path evolution characteristic sequence at the current moment.

[0126] During each simulation time interval, the dynamic simulation module reads information from the load transfer path evolution characteristic sequence at the current moment. Based on this information, it determines the direction of load transfer within the pile foundation from one structural unit to another, and the direction of load transfer from the pile foundation structural unit to adjacent soil layer units.

[0127] At the same time, determine the proportion of load transfer along different transfer paths. For example, what proportion of the total load is transferred downwards along the pile shaft to the pile tip, and what proportion is transferred through the pile side to the surrounding soil layer units.

[0128] Step S143: Determine the action type and intensity of the pile-soil interface based on the current pile-soil interface state evolution characteristic sequence, and adjust the load transfer efficiency according to the action type and intensity.

[0129] Based on the current sequence of pile-soil interface state evolution characteristics, determine the type of action at the pile-soil interface, such as whether pile side friction or pile tip normal force is the primary force at this moment. Simultaneously, determine the intensity of the action, i.e., the magnitude of the friction and normal forces. Adjust the load transfer efficiency according to the type and intensity of the action; for example, higher intensity results in higher load transfer efficiency, meaning more load can be transferred through the interface; lower intensity results in lower load transfer efficiency.

[0130] Step S144: Calculate the load increase at each location of the pile foundation and the stress increase of each soil unit based on the transmission direction, the transmission ratio, and the load transmission efficiency.

[0131] Based on the direction of load transfer, determine the flow path of the load in the pile foundation and soil; according to the transfer ratio, distribute the total load within the time interval to each transfer path; and then, in combination with the load transfer efficiency, calculate the actual effective load transfer amount on each transfer path.

[0132] For each location of the pile foundation, calculate the difference between the load received and the load transmitted during the time interval to obtain the load increase; for each soil unit, calculate the stress increase based on the effective load transmitted to that unit and the soil properties.

[0133] Step S1441: Determine the load distribution path between the pile foundation structural unit and the soil layer unit based on the transmission direction, and generate a load distribution path diagram.

[0134] Based on the direction of load transfer, the load distribution path between the pile foundation structural unit and the soil layer unit is plotted, forming a load distribution path diagram. In the diagram, lines represent the direction of load transfer, with the starting point of the line being the unit from which the load flows out and the ending point being the unit into which the load flows in. The above diagram visually demonstrates the load distribution path.

[0135] Step S1442: Distribute the total load to each branch path in the load distribution path diagram according to the transmission ratio, and obtain the load distribution value of each branch path.

[0136] According to the transfer ratio, the total load value within the simulation time interval is distributed to each branch path in the load distribution path diagram. For example, if the transfer ratio of a certain branch path is a certain ratio, then the load distribution value of that branch path is equal to the total load value multiplied by that ratio, and so on, to obtain the load distribution values ​​of all branch paths.

[0137] Step S1443: Determine the load transfer efficiency coefficient of each branch path based on the action intensity parameter in the pile-soil interface state evolution characteristic sequence. The load transfer efficiency coefficient is positively correlated with the action intensity parameter.

[0138] Based on the intensity parameters, such as the magnitude of friction and normal force, in the pile-soil interface state evolution sequence, the load transfer efficiency coefficient for each branch path is determined. A larger intensity parameter corresponds to a larger load transfer efficiency coefficient, showing a positive correlation. For example, a certain intensity parameter corresponds to a certain load transfer efficiency coefficient.

[0139] Step S1444: Multiply the load distribution value of each branch path by the corresponding load transfer efficiency coefficient to obtain the effective load transfer value of each branch path.

[0140] Multiply the load distribution value of each branch path by its corresponding load transfer efficiency coefficient to obtain the effective load transfer value of that branch path within that time interval. This effective load transfer value represents the actual amount of load that can be transferred through that path.

[0141] Step S1445: Calculate the load increase of each structural unit of the pile foundation based on the effective load transfer value. The load increase is equal to the effective load transfer value flowing into the structural unit minus the effective load transfer value flowing out of the structural unit.

[0142] For each structural unit of the pile foundation, sum up all the effective load transfer values ​​flowing into the unit, and then subtract all the effective load transfer values ​​flowing out of the unit. The difference is the load increase of the structural unit during the time interval.

[0143] Step S1446: Calculate the stress increase of each soil element based on the effective load transfer value and the stiffness parameter of the soil layer element. The stress increase is positively correlated with the effective load transfer value and negatively correlated with the stiffness parameter.

[0144] The stress increase is calculated based on the effective load transfer value and the stiffness parameters of the soil layer elements. The larger the effective load transfer value, the greater the stress increase; the larger the stiffness parameters of the soil layer elements, the smaller the stress increase. By analyzing the relationship between these two factors, the stress increase of each soil element within that time interval can be calculated.

[0145] Step S145: The load increase and stress increase are superimposed with the load and stress values ​​of the previous moment to obtain the load distribution state and stress distribution state at the current moment.

[0146] The calculated load increase at each location of the pile foundation is added to the load value at that location at the previous moment to obtain the load value at each location of the pile foundation at the current moment, and the load distribution state at the current moment is formed by combining them; the stress increase at each element of the soil is added to the stress value of that element at the previous moment to obtain the stress value of each element of the soil at the current moment, and the stress distribution state at the current moment is formed by combining them.

[0147] Step S146: Record the load distribution state and stress distribution state for each simulation time interval in chronological order, extract the characteristics of the load distribution state and stress distribution state changing over time, and generate dynamic characteristics of pile foundation load distribution and pile displacement evolution characteristics. The dynamic characteristics of pile foundation load distribution and the pile displacement evolution characteristics include load evolution trend and displacement evolution trend information.

[0148] Record the load distribution and stress distribution at the end of each simulation time interval in chronological order. Analyze the changes in these states over time and extract the characteristics of these changes, such as the increasing and decreasing trends of load at different locations in the pile foundation and the diffusion trends of stress in different soil elements.

[0149] Based on the changing characteristics of the load distribution state, the dynamic characteristics of the pile foundation load distribution are generated. Based on the stress distribution state and the deformation characteristics of the soil and pile foundation, the displacement change of the pile body is analyzed, the displacement change trend is extracted, and the pile body displacement evolution characteristics are generated.

[0150] Step S150: Generate a pile foundation load transfer simulation analysis report containing load transfer dynamic process curves based on the dynamic characteristics of the pile foundation load distribution and the pile body displacement evolution characteristics.

[0151] By comprehensively considering the dynamic characteristics of pile foundation load distribution and the evolution characteristics of pile displacement, these characteristics are presented in an appropriate form, including plotting load transfer dynamic process curves, and then combining textual descriptions and analysis to generate a complete pile foundation load transfer simulation analysis report.

[0152] Step S151: Extract the evolution information of the load transfer path from the dynamic characteristics of the pile foundation load distribution and the evolution characteristics of the pile body displacement. The evolution information includes the transfer direction, transfer coverage and transfer intensity information at different time points.

[0153] By analyzing the dynamic characteristics of pile foundation load distribution, the load distribution within the pile foundation at different time points is analyzed to infer the load transfer direction and coverage area; combined with the pile displacement evolution characteristics, the accuracy of the transfer direction is further verified.

[0154] Simultaneously, based on the density and magnitude of the load distribution, the load transfer intensity information at different time points is determined. This information on transfer direction, coverage area, and intensity is then compiled to form the evolution information of the load transfer path.

[0155] For example, step S1511: perform data analysis processing on the dynamic characteristics of the pile foundation load distribution and the characteristics of the pile body displacement evolution to separate the characteristic components related to the load transfer path.

[0156] Data analysis and processing are performed on the dynamic characteristics of pile foundation load distribution and pile displacement evolution characteristics to remove interference information unrelated to the load transfer path, such as some local and short-term load fluctuations or displacement fluctuations.

[0157] Retain and extract characteristic components directly related to the load transfer path, such as the distribution changes of the load in the main transfer direction and the pile displacement changes corresponding to the transfer path.

[0158] Step S1512: Extract the transmission direction parameters for all time points from the feature components. The transmission direction parameters are used to represent the transmission angle of the load in space.

[0159] From the separated characteristic components, for each time point, the main direction of load transfer is determined and represented by a transfer direction parameter, which is an angle value that can accurately reflect the transfer angle of the load in space.

[0160] Step S1513: Extract the load transfer coverage parameters for all time points. The load transfer coverage parameters are used to represent the range of pile foundation length and soil depth covered by the load transfer.

[0161] For each time point, the load transfer coverage parameter is extracted. This load transfer coverage parameter includes the length range of the pile foundation involved in the load transfer, that is, the length interval from the top of the pile to a certain position in the pile body, and the depth range of the soil involved, that is, the interval from the ground surface to a certain depth of the soil. These parameters clearly represent the load transfer coverage.

[0162] Step S1514: Extract the transfer intensity parameters at all time points, whereby the transfer intensity parameters are used to represent the magnitude of load transfer.

[0163] At each time point, based on the dynamic characteristics of the load distribution, the transfer strength parameter is extracted. This transfer strength parameter can be the average load size on the transfer path at that time point, or the maximum load value, etc., to represent the magnitude of load transfer.

[0164] Step S1515: Pair the transmission direction parameter, the transmission coverage parameter, and the transmission intensity parameter according to time points to form an evolution data record containing time stamps, transmission direction parameters, transmission coverage parameters, and transmission intensity parameters.

[0165] The transmission direction parameters, transmission coverage parameters, and transmission intensity parameters at each time point are paired with the corresponding time markers to form an evolution data record. Each record fully contains all the characteristic parameters of the load transmission path at that time point.

[0166] Step S1516: Sort the evolution data records and arrange them in chronological order to generate a complete evolution dataset. The evolution dataset is used to plot the dynamic process curve of load transfer.

[0167] All evolutionary data records were sorted according to their time stamps to form a complete evolutionary dataset. This dataset clearly presents the characteristics of the load transfer path at each time point in chronological order.

[0168] Step S152: Input the evolution information into the curve generation component and draw a dynamic process curve of the load transfer path changing over time. The horizontal axis of the dynamic process curve represents time, the vertical axis represents the transfer intensity, and the shape of the curve represents the changes in the transfer direction and the transfer coverage.

[0169] The evolutionary information is input into the curve generation component. The component draws a basic curve framework based on the time stamps and transmission intensity parameters in the evolutionary dataset, with time on the horizontal axis and transmission intensity on the vertical axis.

[0170] At the same time, changes in the shape of the curve can be used to represent changes in the transmission direction and coverage area. For example, the slope of the curve indicates a change in the transmission direction, and the width of the curve indicates the size of the transmission coverage area. The wider the curve, the wider the coverage area.

[0171] Step S153: Extract the change information of the pile-soil interface action state from the dynamic characteristics of the pile foundation load distribution and the pile displacement evolution characteristics, and generate a state change time series table in chronological order. The state change time series table includes the action type, action intensity and corresponding load transfer characteristics of each time interval.

[0172] Extract the type of action of the pile-soil interface in different time intervals from the dynamic characteristics of pile foundation load distribution. For example, in a certain time interval, the pile-soil interface mainly exhibits frictional force, while in another time interval, frictional force and normal force may coexist.

[0173] Simultaneously, the intensity of the load is extracted within each time interval. This intensity is reflected by the magnitude of the force transmitted at the pile-soil interface, which can be determined based on the interaction force data between the pile structural units and soil layer units in the dynamic characteristics of pile foundation load distribution. Furthermore, by combining the pile displacement evolution characteristics, the corresponding load transmission characteristics within each time interval are determined, such as the main direction of load transmission and the transmission range.

[0174] The extracted action types, action intensities, and corresponding load transfer characteristics are arranged in chronological order to generate a state change time series table. Each row in the table corresponds to relevant information for a time interval, so that the changes in the action state of the pile-soil interface can be clearly presented.

[0175] Step S154: Visualize the dynamic process curve and the state change time series table by adding titles, axis labels and legends to generate a visualization chart.

[0176] When visualizing dynamic process curves, add appropriate titles to the curves. The titles should accurately reflect the content represented by the curves, such as "Dynamic process curve of pile foundation load transfer path changing over time".

[0177] Add a label “Time” to the horizontal axis and indicate the unit of time, such as “second”; add a label “Transmission Intensity” to the vertical axis and indicate its corresponding dimensionless unit.

[0178] Add legends to explain the transmission direction and coverage area represented by the curve shape. For example, use the direction indicators corresponding to different curve inclination angles to explain the transmission direction, and use the range indicators corresponding to the curve width to explain the transmission coverage area.

[0179] When visualizing the state change time series table, add a title to the table "Pile-Soil Interface Action State Change Time Series Table" and clearly label the column headings in the table, such as "Time Interval", "Action Type", "Action Intensity", "Load Transfer Characteristics", etc.

[0180] These visualization processes make dynamic process curves and state change time series tables clearer and easier to understand, forming complete visualization charts.

[0181] Step S155: Integrate the visualization charts with the textual descriptions of the dynamic characteristics of pile foundation load distribution and the evolution characteristics of pile displacement, format them according to the preset report format, and generate a pile foundation load transfer simulation analysis report containing the dynamic process curve of load transfer.

[0182] Collect textual descriptions of the dynamic characteristics of pile foundation load distribution. These descriptions include the pattern of load distribution across each structural unit of the pile foundation over time, the location and time of the maximum load value, and so on.

[0183] Collect textual descriptions of the pile displacement evolution characteristics, including the trend of displacement changes of various parts of the pile over time, the location and time of the maximum displacement value, etc.

[0184] Integrate these textual descriptions with the previously generated visualizations and format them according to a preset report format. The report format typically includes an abstract, introduction, simulation model introduction, load transfer process analysis, results visualization, and conclusions.

[0185] During the layout process, dynamic process curves and state change time series tables are inserted into the corresponding analysis sections, and textual descriptions are provided in detail around the charts, enabling readers to fully understand the simulation results of pile foundation load transfer through a combination of text and charts.

[0186] Finally, a complete pile foundation load transfer simulation analysis report is generated, which includes load transfer dynamic process curves.

[0187] Furthermore, this embodiment may also include the step of training a dynamic evolution model of load transfer.

[0188] In order for the load transfer dynamic evolution model to accurately process the set of load transfer process segments and generate the corresponding evolution feature sequence, the load transfer dynamic evolution model needs to be trained. The specific steps are as follows.

[0189] Step S211: Collect a large amount of sample data on the interaction between pile foundation and soil. The sample data includes records of load transfer processes under different pile foundation types, different soil environments, and different load conditions.

[0190] A large amount of sample data was collected through methods such as reviewing engineering archives, laboratory tests, and numerical simulations. The pile foundation types in the sample data include pile foundations of different materials, sizes, and structural forms, such as reinforced concrete piles, wooden piles, and steel pipe piles; the different soil environments cover various soil layer combinations and situations with significant differences in soil physical and mechanical properties; and the different load conditions include various load magnitudes, application methods, and variations.

[0191] Each sample data contains a complete record of the load transfer process, such as the load transfer situation of each part of the pile foundation and the stress response of each soil unit, as mentioned above, to ensure that the sample data is diverse and representative and can cover a variety of possible engineering scenarios.

[0192] Step S212: Preprocess the collected sample data, including data cleaning, data standardization, and data partitioning.

[0193] Each sample data record is examined one by one to identify noisy data, which may be abnormal fluctuations caused by measurement equipment errors or data recording errors. Noise is removed through smoothing and other methods.

[0194] For abnormal values ​​that deviate significantly from the normal range, such as load values ​​at a certain time point being much greater than load values ​​at other time points and not conforming to the load change pattern, they should be removed.

[0195] For data records with missing values, interpolation is performed to fill in the missing values ​​based on the trend of the data before and after, ensuring the integrity and continuity of the sample data.

[0196] The sample data contains a variety of characteristic parameters, such as load magnitude (unit: force), displacement (unit: length), and stress (unit: pressure), which have different dimensions.

[0197] The aforementioned characteristic parameters are processed using a standardization method, converting each parameter into a dimensionless numerical value to ensure comparability between different parameters. For example, the load magnitude is converted into a proportion relative to the maximum load value in the sample data; the displacement is converted into a proportion relative to the pile foundation length, and so on.

[0198] Then, according to a preset ratio, such as 70%, 15%, and 15%, the standardized sample data is randomly divided into a training set, a validation set, and a test set. The training set is used for parameter learning and training of the model; the validation set is used to evaluate the model's performance during training and adjust the model's hyperparameters, such as the learning rate and the number of network layers; the test set is used to evaluate the model's generalization ability after training is completed and to test the model's performance on unseen data.

[0199] Step S213: Construct the network structure of the load transfer dynamic evolution model, including modules such as process coding component, path tracing component, and state association component, and set the initial parameters of each module.

[0200] A process coding component is constructed using a recurrent neural network structure, which can effectively process time series data. Initial parameters such as the number of hidden layers and the number of neurons in each hidden layer are set to enable it to effectively encode the time series data of the load transfer process.

[0201] A path tracing component is constructed using a convolutional neural network structure. By setting convolutional kernels of different sizes, local features in the process-encoded data sequence are extracted, thereby capturing the features of the load transfer path. Initial parameters such as the number of convolutional layers and the size of the convolutional kernels are set.

[0202] A state association component is constructed using a graph neural network structure, which is suitable for processing data with correlations. Initial parameters such as the feature dimension of the graph nodes and the connection method of the edges are set to analyze the correlation between the interaction intensity of the pile-soil interface and the load transfer path.

[0203] At the same time, initial parameters such as the optimizer type and loss function of the model are set. The optimizer is used to update the model parameters, and the loss function is used to measure the difference between the model's prediction results and the actual results.

[0204] Step S214: Input the training set into the constructed load transfer dynamic evolution model for training, and continuously adjust the model parameters through the backpropagation algorithm.

[0205] The sample data in the training set are input into the load transfer dynamic evolution model in batches. The model processes the input data and generates prediction results of the load transfer path evolution feature sequence and the pile-soil interface state evolution feature sequence.

[0206] The predicted results are compared with the actual results in the sample data, the value of the loss function is calculated, and the parameters of each module, such as the weights of the recurrent neural network and the convolution kernel parameters of the convolutional neural network, are adjusted backward from the output layer to the input layer based on the value of the loss function through the backpropagation algorithm.

[0207] Repeat the above process, iterating continuously until the value of the loss function reaches the preset threshold or the training reaches the maximum number of iterations.

[0208] Step S215: Use the validation set to evaluate the model during training and adjust the model's hyperparameters based on the evaluation results.

[0209] After each iteration of model training, the validation set is input into the current model to obtain the validation results.

[0210] The performance of the model on the validation set is evaluated by calculating error metrics between the validation results and the actual results on the validation set, such as mean absolute error and mean squared error.

[0211] Based on the evaluation results, adjust the model's hyperparameters, such as increasing or decreasing the number of neurons in the hidden layer and adjusting the learning rate, to optimize the model's performance on the validation set.

[0212] Step S216: After the model training is completed, use the test set to perform a final evaluation of the model to check its generalization ability. If the evaluation results meet the preset requirements, the model training is complete.

[0213] The test set is input into the trained model to obtain the test results, and the error index between the test results and the actual test set results is calculated.

[0214] If the error index is within the preset acceptable range, it indicates that the model has good generalization ability and can accurately analyze and predict unseen load transfer processes, and the model training is complete.

[0215] If the error index exceeds the preset range, the sample data needs to be re-examined or the model structure adjusted, and the training and evaluation should be repeated until the model meets the preset requirements.

[0216] To ensure the accuracy and reliability of the pile foundation load transfer simulation analysis report, it needs to be verified. The specific steps are as follows.

[0217] Step S311: Select an actual engineering case similar to the simulation scenario and collect the measured data of the actual engineering case, including the load distribution data and pile displacement data of the pile foundation under actual load.

[0218] Through research, we selected actual engineering cases with similar pile foundation type, soil environment and load conditions to those in this embodiment. These actual engineering cases should have complete monitoring data.

[0219] Collect measured data obtained through monitoring equipment during the construction and use of this project, such as load change data at the pile top, stress monitoring data at different depths of the pile body, displacement monitoring data of the pile body, and stress and deformation data of the surrounding soil.

[0220] Ensure that the collected measured data has a sufficient time span and data accuracy to reflect the actual load transfer process and the response of the pile foundation and soil.

[0221] Step S312: Compare and analyze the prediction results in the simulation analysis report with the measured data of the actual engineering case, and calculate the deviation between the two.

[0222] Extract key prediction results from the simulation analysis report, such as the load distribution of the pile foundation at different time points and the evolution characteristics of pile displacement.

[0223] These predictions are compared with measured data from actual engineering cases at the same time and location. The deviation value for each comparison point is calculated, and the deviation value is the difference between the prediction result and the measured data.

[0224] Statistical analysis was conducted on the deviation values ​​of all comparison points, and statistical indicators such as average deviation and maximum deviation were calculated to gain a comprehensive understanding of the differences between simulation results and measured data.

[0225] Step S313: Based on the deviation analysis results, evaluate the accuracy of the simulation analysis report. If the deviation is within an acceptable range, the simulation analysis report is deemed valid; if the deviation exceeds the acceptable range, return to the corresponding step for correction.

[0226] Based on the requirements for the accuracy of simulation results in engineering practice, an acceptable range of deviation is set.

[0227] Compare the calculated average deviation, maximum deviation, and other statistical indicators with the acceptable range. If all indicators are within the acceptable range, it means that the simulation analysis report can accurately reflect the actual pile foundation load transfer situation, and the report is valid.

[0228] If any indicators exceed the acceptable range, the cause of the deviation may be unreasonable parameter settings in the interaction scenario model, insufficient training of the load transfer dynamic evolution model, or improper parameter adjustment during the dynamic simulation process.

[0229] Based on the cause of the deviation, return to the corresponding steps for correction, such as readjusting the soil parameters of the interaction scenario model, retraining the load transfer dynamic evolution model, or optimizing the parameters of the dynamic deduction. After the correction is completed, regenerate the simulation analysis report and verify it again until the report meets the accuracy requirements.

[0230] Therefore, the verified pile foundation load transfer simulation analysis report can be stored and archived, and an index of the report can be established for subsequent querying and retrieval.

[0231] For example, after a simulation analysis report passes verification, it needs to be stored and archived. First, a unique identifier is assigned to the report, which includes information such as the date the report was generated, the corresponding load condition type, and the model version, in order to distinguish different reports.

[0232] Next, the electronic document of the report is stored in a designated database or file server, with the storage path set according to the established classification rules, such as categorizing by project name, analysis time, etc. Simultaneously, an index is created for the report, including its identifier, generation time, overview of load conditions, and main conclusions. This index information will be stored in an index database, allowing users to easily search for and quickly locate the required report using keywords.

[0233] In addition, stored reports are backed up regularly to prevent data loss. The backup frequency is determined based on the importance and update frequency of the reports. Backup files are also managed according to the same classification rules and index information to ensure that report data can be restored promptly when needed.

[0234] This process also allows for periodic review and summarization of stored pile foundation load transfer simulation analysis reports, extracting common patterns and special cases. For example, periodic review and summarization of stored simulation analysis reports can be performed, such as periodically organizing all reports generated within a given timeframe. During the review, the characteristics of pile foundation load transfer under different load conditions can be compared, analyzing common patterns. For instance, in the same type of soil, are there similarities in the evolution trends of pile foundation load transfer paths? Do the changes in the pile-soil interface state follow certain patterns?

[0235] At the same time, we should pay attention to special cases, namely those cases where the load transfer characteristics are significantly different from the usual situation, and analyze the reasons for the differences, such as special soil distribution and complex load change forms.

[0236] The extracted common patterns and special cases are compiled into a report summary document, which details the specific manifestations of the patterns, the analysis of special cases, and the corresponding conclusions. These summary documents will serve as reference materials for subsequent pile foundation design and analysis, helping designers to draw on previous simulation analysis results, optimize design schemes, and improve the rationality and reliability of designs when designing similar projects.

[0237] Furthermore, the dynamic evolution model of load transfer is regularly maintained and updated to adapt to different engineering scenarios and new technical requirements.

[0238] To ensure that the load transfer dynamic evolution model can be continuously and effectively applied to different engineering scenarios and meet new technical requirements, it needs to be maintained and updated regularly. First, a maintenance and update cycle should be established, which can be determined based on the model's usage frequency, the rate of change in engineering scenarios, and the progress of technological development.

[0239] During maintenance, check whether each component of the model is operating normally, such as whether the process coding component, path tracing component, and state association component are functioning correctly, and whether the accuracy of data processing meets the requirements. If any abnormalities or data processing errors are found in the components, repair and debug them in a timely manner, such as correcting algorithmic logic errors in the components and optimizing the data processing flow.

[0240] The main aspects of model updates include: First, adjusting model parameter settings, such as the size of the path recognition window and thresholds in correlation analysis, based on new engineering practice data and research results, to improve the model's adaptability to new scenarios; Second, introducing new feature parameters or improving existing feature extraction methods. For example, when certain new soil parameters are found to have a significant impact on load transfer, they are included in the model's analysis scope, and the processing logic of the process coding component and state correlation component is adjusted accordingly; Third, optimizing the model's structure, such as increasing the number of neural network layers or adjusting the number of neurons in each layer, to improve the model's analytical accuracy and ability to process complex data.

[0241] After maintaining and updating the model, it needs to be tested. A new test dataset should be used to verify whether the model's performance meets the requirements. Testing includes assessing the model's analytical accuracy and operational efficiency. Only models that pass the tests can be deployed. The updated model version information should be recorded in the model management document, and simulation analysis reports generated based on the old version should be marked for easy differentiation.

[0242] Furthermore, new pile foundation engineering case data and load transfer test data can be collected and used as new training samples to optimize and improve the load transfer dynamic evolution model.

[0243] We continuously collect new data on pile foundation engineering cases and load transfer tests. This data includes the structural parameters of pile foundations in actual projects, the physical and mechanical properties of the soil, the applied load conditions, and the stress and deformation monitoring data of pile foundations and soil, as well as detailed data obtained from pile foundation load transfer tests conducted in the laboratory, such as the stress distribution of the pile body and the changes in friction at the pile-soil interface at different load stages.

[0244] The collected data undergoes preprocessing. First, outliers and erroneous data are removed. For example, by comparing the data with the reasonable range of similar data, data that deviates significantly is identified and corrected or deleted. Then, the data is standardized, converting parameters with different dimensions into dimensionless parameters. For example, parameters such as load magnitude and soil weight are scaled according to a set ratio to bring them within the same numerical range, thus meeting the training requirements of the model.

[0245] The preprocessed new data is used as new training samples and merged with the original training samples to form an updated training dataset. During the merging process, it is ensured that the format and feature parameters of the new samples are consistent with those of the original samples to facilitate unified model training. Simultaneously, the training dataset is divided: one part is used as the training set for model parameter optimization, and the other part is used as the validation set to evaluate the model's improvement.

[0246] Based on the new training samples, the load transfer dynamic evolution model is retrained using incremental training to improve the model's generalization ability and analysis accuracy.

[0247] Incremental training is used to retrain the load transfer dynamic evolution model. This means that the model is further trained using new training samples based on the already trained model parameters, rather than retraining the entire model. First, the parameters of the currently used model are loaded and used as the initial parameters for retraining.

[0248] Next, set the parameters for incremental training, such as the learning rate and the number of training iterations. The learning rate can be set slightly lower than the initial learning rate to avoid excessive interference with the already learned knowledge. The number of training iterations is determined based on the number of new training samples and the model's convergence.

[0249] During training, new training samples are input into the model, and the model adjusts the weight parameters of each layer using backpropagation, allowing the model to gradually adapt to the information contained in the new samples. Simultaneously, in each iteration, the model's performance is evaluated using a validation set, and the prediction error is calculated. Training stops when the prediction error reaches a preset threshold or the number of training iterations reaches a set value.

[0250] After training is complete, save the updated model parameters and perform performance tests on the model. Compare the model's analysis results on the test dataset before and after the update to evaluate whether the model's generalization ability and analytical accuracy have improved. If the model performance meets the requirements, it is used as a new version; if it does not meet the requirements, the incremental training parameters are adjusted or the number of training samples is increased, and training is performed again.

[0251] Establish a model performance evaluation index system, and regularly evaluate the performance of the load transfer dynamic evolution model based on this index system to ensure the effectiveness and reliability of the load transfer dynamic evolution model.

[0252] A model performance evaluation index system is established, which includes multiple evaluation indicators, such as prediction error rate, which is the proportion of error between the model's predicted pile foundation load distribution and displacement evolution results and the actual monitoring data or experimental data; model stability, which is the consistency of the output results when the model runs the same input data multiple times; computational efficiency, which is the time required for the model to process a certain amount of data; and generalization ability, which is the accuracy of the model in analyzing new data that was not used in the training.

[0253] For each evaluation metric, set corresponding evaluation criteria and thresholds, such as the acceptable threshold for prediction error rate and the minimum requirement for model stability. These criteria and thresholds are determined based on the needs of engineering practice and the application scenario of the model.

[0254] The performance of the load transfer dynamic evolution model is periodically evaluated based on this index system, with the evaluation cycle consistent with the model's maintenance and update cycle. During the evaluation process, a representative test dataset is selected, input into the model, and the model's output results are obtained. Then, according to the calculation methods of each evaluation index, the index values ​​of the model on the test dataset are calculated.

[0255] The calculated index values ​​are compared with the preset thresholds. If all indexes meet the requirements, it indicates that the model performs well and can maintain effectiveness and reliability. If any index does not meet the requirements, the reasons for the index failure are analyzed, such as unreasonable model parameter settings or insufficient training samples. Corresponding improvement measures are taken for the reasons, such as retraining the model or supplementing training samples, until the model performance reaches the evaluation standard.

[0256] Therefore, based on the generated pile foundation load transfer simulation analysis report, suggestions can be provided for pile foundation design optimization, including pile foundation size adjustment, material selection, and construction process improvement.

[0257] Based on the analysis results of the stress and deformation characteristics of the pile foundation under different load conditions in the pile foundation load transfer simulation analysis report, specific suggestions are provided for the design optimization of the pile foundation. Regarding pile foundation size adjustment, if the report shows that under a certain load, the load on a certain part of the pile is too concentrated, exceeding the material's bearing capacity, it can be recommended to increase the cross-sectional dimensions of that part to improve its bearing capacity; if insufficient pile length leads to excessive stress on the pile tip, it can be recommended to appropriately increase the pile length so that the load can be transferred more evenly to the deeper soil.

[0258] Regarding material selection, based on the requirements for pile foundation material performance in the report, if the existing material has a low elastic modulus, resulting in excessive pile deformation, it is recommended to select a material with a higher elastic modulus. If the friction at the pile-soil interface is insufficient, affecting the load transfer efficiency, it is recommended to treat the pile surface by using a material with a higher coefficient of friction or a material with increased surface roughness.

[0259] Regarding improvements in construction techniques, if the report indicates that the load transfer path is abnormal due to soil disturbance during pile foundation construction, it is recommended to optimize the construction techniques, such as using more advanced pile forming techniques to reduce disturbance to the surrounding soil. If the bond between the pile body and the soil is not tight enough, affecting the interfacial strength, it is recommended to improve the grouting process to enhance the bond between the pile body and the soil.

[0260] These suggestions are compiled into a design optimization report, which details the basis for the suggestions, namely the relevant data and conclusions in the simulation analysis report, as well as the potential benefits of implementing the suggestions, such as improving the bearing capacity of the pile foundation, reducing deformation, and reducing engineering costs, thus providing designers with a clear direction for optimization.

[0261] Based on the analysis of the pile-soil interface state in the simulation analysis report, the long-term stability of the pile foundation is evaluated, and the performance change trend of the pile foundation during long-term use is predicted.

[0262] Based on the analysis results of the pile-soil interface interaction state changing over time in the simulation analysis report, the long-term stability of the pile foundation is evaluated. The attenuation of the pile-soil interface interaction strength under long-term load is analyzed, along with the impact of this attenuation on the load transfer path and the stress state of the pile foundation.

[0263] By combining the creep characteristics of soil and the fatigue properties of pile foundation materials, the performance change trend of pile foundations during long-term use can be predicted, such as the cumulative amount of pile displacement over time and the long-term variation law of pile-soil interface friction.

[0264] If the prediction results indicate that the pile foundation may experience excessive deformation or a decrease in bearing capacity during long-term use, exceeding the allowable range of the project, corresponding maintenance and reinforcement suggestions should be made, such as regularly monitoring the pile foundation and taking measures such as grouting reinforcement to enhance the strength of the pile-soil interface, so as to ensure the long-term stable operation of the pile foundation.

[0265] Finally, the simulation analysis of pile foundation load transfer is compared and analyzed with actual engineering monitoring data to continuously improve the simulation model and analysis methods, thereby enhancing the scientificity and reliability of pile foundation engineering design and construction.

[0266] Collect actual monitoring data during the construction and use of pile foundation projects. The actual monitoring data includes measured values ​​of pile settlement, pile stress, pile top load, etc. The frequency and time span of the monitoring data collection should match the time range of the simulation analysis.

[0267] The actual monitoring data is compared and analyzed with the predicted results in the corresponding pile foundation load transfer simulation analysis report. Differences between the two are calculated, such as deviations in settlement and stress distribution. The reasons for these differences may include discrepancies between the soil parameter values ​​in the simulation model and the actual soil, inconsistencies between the simulated load conditions and actual conditions, and the model's simplification neglecting certain influencing factors. Based on the results of the difference analysis, the simulation model and analysis methods are improved and refined. For example, the physical and mechanical parameters of the soil in the model are adjusted to be closer to actual values, the simulation method of load conditions is optimized, and the consideration of neglected factors in the model is increased. By continuously comparing and providing feedback between simulation analysis and actual monitoring data, the accuracy of the simulation model and the rationality of the analysis methods are gradually improved, thereby enhancing the scientific rigor and reliability of pile foundation engineering design and construction, and reducing engineering risks.

[0268] Figure 2The illustration shows exemplary hardware and software components of a pile foundation load transfer simulation system 100 incorporating machine learning, which can implement the ideas of this application, according to some embodiments of this application. For example, a processor 120 may be used in the pile foundation load transfer simulation system 100 incorporating machine learning and to perform the functions described in this application.

[0269] For example, a pile foundation load transfer simulation system 100 incorporating machine learning may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and various forms of storage media 140, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the pile foundation load transfer simulation system 100 incorporating machine learning may also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The methods of this application can be implemented according to these program instructions. The pile foundation load transfer simulation system 100 incorporating machine learning also includes an I / O interface 150 between the computer and other input / output devices.

[0270] Furthermore, this embodiment of the invention also provides a readable storage medium, which has computer-executable instructions pre-set in it. When the processor executes the computer-executable instructions, the above-mentioned pile foundation load transfer simulation method combined with machine learning is realized.

[0271] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A pile foundation load transfer simulation method combining machine learning, characterized in that, The method includes: Construct an interaction scenario model between the pile foundation and the soil, wherein the interaction scenario model includes pile foundation structural elements, soil layer elements, and external load elements; Based on the interaction scenario model, a set of load transfer process segments is generated, which contains continuous process records of the interaction between the pile foundation and the soil under different load conditions. The pre-trained load transfer dynamic evolution model is invoked to perform inversion learning on the set of load transfer process segments, generating load transfer path evolution feature sequences and pile-soil interface state evolution feature sequences. Based on the load transfer path evolution characteristic sequence and the pile-soil interface state evolution characteristic sequence, dynamic deduction of load transfer is performed to obtain the dynamic characteristics of pile foundation load distribution and pile displacement evolution characteristics. Based on the dynamic characteristics of the pile foundation load distribution and the evolution characteristics of the pile body displacement, a pile foundation load transfer simulation analysis report containing load transfer dynamic process curves is generated.

2. The pile foundation load transfer simulation method combined with machine learning according to claim 1, characterized in that, The constructed interaction scenario model between the pile foundation and the soil includes: Determine the spatial boundary of the interaction between the pile foundation and the soil, wherein the spatial boundary includes the spatial region where the pile foundation is located and the spatial region affected by the soil; Within the spatial boundary, pile foundation structural units and soil layer units are divided. The pile foundation structural units correspond to the physical components of the pile foundation, and the soil layer units correspond to the physical components of different soil layers. The interaction rules between the pile foundation structural unit and the soil layer unit are defined. The interaction rules are used to describe the specific ways in which the pile foundation interacts with the surrounding soil when it is under load. Configure the initial state parameters of the pile foundation structural unit and the initial state parameters of the soil layer unit, wherein the initial state parameters include the initial stress and initial displacement conditions; The effectiveness of the interaction scenario model is verified to ensure that the spatial boundary, the pile foundation structural unit, the soil layer unit, the interaction mode rules, and the initial state parameters are mutually compatible.

3. The pile foundation load transfer simulation method combining machine learning according to claim 1, characterized in that, The generation of a set of load transfer process fragments based on the interaction scenario model includes: In the interaction scenario model, various load conditions are set, including different load application locations, different load application methods, and different load variation forms; For each of the aforementioned load conditions, the interaction scenario model is run to simulate the process and record all information about the pile foundation from the start of being stressed to reaching a stable state. The information about the entire process includes the load transfer situation of each part of the pile foundation and the stress response situation of each unit of the soil. The entire process information is segmented according to time sequence, and the entire process information is divided into multiple continuous process segments, each of which corresponds to a specific stage in the load transfer process. Identify key change nodes in each process segment, including nodes where the load transfer direction changes and nodes where the pile-soil interface interaction intensity changes. The process segments containing the key change nodes are classified and organized according to the load application conditions to form a set of load transfer process segments with temporal sequence correlation. Each process segment in the set of load transfer process segments contains complete load transfer stage characteristics.

4. The pile foundation load transfer simulation method combining machine learning according to claim 1, characterized in that, The pre-trained load transfer dynamic evolution model is invoked to perform inversion learning on the set of load transfer process segments, generating a load transfer path evolution feature sequence and a pile-soil interface state evolution feature sequence, including: The load transfer process segment set is input into the process coding component of the load transfer dynamic evolution model to encode and convert the time series data in the load transfer process segment set to generate a process coding data sequence, which is used to represent the time series characteristics of the load transfer process. The path tracking component of the load transfer dynamic evolution model captures the path features of the process-encoded data sequence, analyzes the changes in the load transfer path inside the pile foundation and between the pile foundation and the soil, and generates the initial features of the load transfer path. The state association component of the load transfer dynamic evolution model is used to perform association analysis on key change nodes in the process coded data sequence to determine the association relationship between the pile-soil interface interaction intensity and the load transfer path, and generate an association relationship feature description. Based on the correlation characteristics, the initial characteristics of the load transfer path are dynamically adjusted to obtain a load transfer path evolution characteristic sequence that reflects the influence of pile-soil interaction. Based on the load transfer path evolution feature sequence, the state features of the pile-soil interface at different stages are extracted and arranged in chronological order to generate a pile-soil interface state evolution feature sequence, which includes the interface action type and action intensity features at each stage.

5. The pile foundation load transfer simulation method combined with machine learning according to claim 4, characterized in that, The path tracing component of the load transfer dynamic evolution model captures path features of the process-encoded data sequence, analyzes the changes in the load transfer path within the pile foundation and between the pile foundation and the soil, and generates initial characteristics of the load transfer path, including: A path identification window is set in the path tracing component, which is used to extract local time series segments in the process encoded data sequence; The process-encoded data sequence is traversed through the path identification window, and the path direction of each local time series segment is determined to identify the transmission direction and coverage of the load within that local time series segment. The transmission direction and coverage of adjacent local time series segments are compared and analyzed to analyze the continuation and turning points of the load transmission path and generate path change indicators. The process-encoded data sequence is segmented based on the path change identifier to obtain multiple path segments with stable transmission paths; The load transfer path initial features are extracted from each path segment, including the transfer direction features, transfer coverage features, and duration features, and combined to form the load transfer path initial features. These initial features represent the basic characteristics of the load transfer path at different stages.

6. The pile foundation load transfer simulation method combined with machine learning according to claim 4, characterized in that, The state correlation component of the load transfer dynamic evolution model is used to perform correlation analysis on key change nodes in the process-encoded data sequence to determine the correlation between the pile-soil interface interaction intensity and the load transfer path, and to generate a correlation feature description, including: Extract state description data of all key change nodes from the process-coded data sequence. The state description data includes pile-soil interface interaction strength parameters and load transfer path parameters. The state description data is standardized and transformed, and the standardized state data is input into the correlation analysis model of the state association component to calculate the mutual information value and correlation coefficient value between the pile-soil interface interaction strength parameter and the load transfer path parameter. The degree of correlation between each parameter is determined based on the mutual information value and the correlation coefficient value, and parameter combinations with a degree of correlation exceeding a preset threshold are selected. A relational network structure is constructed based on the selected parameter combinations. In the relational network structure, nodes represent parameters, edges represent the relational relationships between parameters, and the weight of the edges represents the degree of relational tightness. The aforementioned network structure is converted into a relational feature description that includes node attributes and edge attributes. This relational feature description is used to quantify the relationship between the pile-soil interface interaction intensity and the load transfer path.

7. The pile foundation load transfer simulation method combining machine learning according to claim 1, characterized in that, The dynamic deduction of load transfer based on the load transfer path evolution characteristic sequence and the pile-soil interface state evolution characteristic sequence yields the dynamic characteristics of pile foundation load distribution and pile displacement evolution characteristics, including: Input the load transfer path evolution characteristic sequence and the pile-soil interface state evolution characteristic sequence into the dynamic simulation module, initialize the simulation environment parameters, and set the simulation time interval and total simulation duration; Within each simulation time interval, the load transfer direction and transfer ratio in the pile foundation and soil are determined based on the load transfer path evolution characteristic sequence at the current moment. The type and intensity of the action on the pile-soil interface are determined based on the current pile-soil interface state evolution characteristic sequence, and the load transfer efficiency is adjusted according to the type and intensity of the action. Calculate the load increase at each location of the pile foundation and the stress increase of each soil unit based on the transmission direction, the transmission ratio, and the load transmission efficiency. The load increase and stress increase are superimposed with the load and stress values ​​of the previous moment to obtain the load distribution and stress distribution at the current moment. Record the load distribution and stress distribution states for each simulation time interval in chronological order, extract the characteristics of the load distribution and stress distribution states changing over time, and generate dynamic characteristics of pile foundation load distribution and pile displacement evolution characteristics. The dynamic characteristics of pile foundation load distribution and the pile displacement evolution characteristics include load evolution trend and displacement evolution trend information.

8. The pile foundation load transfer simulation method combined with machine learning according to claim 7, characterized in that, The calculation of the load increase at each location of the pile foundation and the stress increase in each soil element based on the transfer direction, the transfer ratio, and the load transfer efficiency includes: Based on the transmission direction, the load distribution path between the pile foundation structural unit and the soil layer unit is determined, and a load distribution path diagram is generated. The total load is distributed to each branch path in the load distribution path diagram according to the said transfer ratio, and the load distribution value of each branch path is obtained. The load transfer efficiency coefficient of each branch path is determined based on the action intensity parameter in the pile-soil interface state evolution characteristic sequence. The load transfer efficiency coefficient is positively correlated with the action intensity parameter. Multiply the load distribution value of each branch path by the corresponding load transfer efficiency coefficient to obtain the effective load transfer value of each branch path. The load increase of each structural unit of the pile foundation is calculated based on the effective load transfer value. The load increase is equal to the effective load transfer value flowing into the structural unit minus the effective load transfer value flowing out of the structural unit. The stress increase of each soil element is calculated based on the effective load transfer value and the stiffness parameters of the soil layer elements. The stress increase is positively correlated with the effective load transfer value and negatively correlated with the stiffness parameters.

9. The pile foundation load transfer simulation method combining machine learning according to claim 1, characterized in that, The pile foundation load transfer simulation analysis report, which generates a dynamic process curve of load transfer based on the dynamic characteristics of the pile foundation load distribution and the evolution characteristics of the pile body displacement, includes: Evolution information of load transfer path is extracted from the dynamic characteristics of the pile foundation load distribution and the characteristics of the pile body displacement. The evolution information includes the transfer direction, transfer coverage and transfer intensity information at different time points. The evolution information is input into the curve generation component to draw a dynamic process curve of the load transfer path changing over time. The horizontal axis of the dynamic process curve represents time, the vertical axis represents the transfer intensity, and the shape of the curve represents the changes in the transfer direction and the transfer coverage. The dynamic characteristics of the pile foundation load distribution and the characteristics of the pile body displacement evolution are used to extract the change information of the pile-soil interface action state. The state change time series table is generated by arranging the state change time series table in chronological order. The state change time series table includes the action type, action intensity and corresponding load transfer characteristics of each time interval. Visualize the dynamic process curve and the state change time series table by adding titles, axis labels and legends to generate visual charts. The visualization charts are integrated with textual descriptions of the dynamic characteristics of pile load distribution and the evolution characteristics of pile displacement. The data is then formatted according to a preset report format to generate a pile load transfer simulation analysis report containing dynamic process curves of load transfer.

10. A pile foundation load transfer simulation system combining machine learning, characterized in that... The device includes a processor and a memory, the memory and the processor being connected. The memory is used to store programs, instructions or code, and the processor is used to execute the programs, instructions or code in the memory to implement the pile foundation load transfer simulation method combined with machine learning as described in any one of claims 1-9.

Images