Machine learning-combined pile foundation load transfer simulation method and system

Abstract

The invention provides a pile foundation load transfer simulation method and system combined with machine learning, and the method comprises the steps: firstly constructing an interaction scene model which comprises a pile foundation construction element, a soil body layer element and an external load element, and generating a load transfer process fragment set under different load conditions; then, a pre-trained load transmission dynamic evolution model is called to conduct inversion learning on the fragment set, a load transmission path and a pile-soil interface state evolution characteristic sequence are generated, dynamic deduction is conducted according to the load transmission path and the pile-soil interface state evolution characteristic sequence, and dynamic characteristics of pile foundation load distribution and pile body displacement are obtained; and finally, a simulation analysis report containing a load transfer dynamic process curve is generated based on the dynamic characteristics, so that the accuracy and reliability of pile foundation load transfer simulation are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of machine learning, in particular to a pile load transfer simulation method and system combined with machine learning. BACKGROUND

[0002] In the field of civil engineering, pile foundation as the foundation support structure of buildings, its load transfer characteristics are crucial to the stability and safety of buildings. Traditional pile load transfer analysis methods mainly rely on theoretical formula derivation and field test. Theoretical formula is often based on some simplifying assumptions, which is difficult to accurately reflect the complex interaction relationship between pile foundation and soil; while field test has the problems of high cost, long cycle, limited by site conditions, and difficult to obtain comprehensive dynamic information of pile load transfer under different load conditions.

[0003] With the development of computer technology, numerical simulation methods have been gradually applied to pile load transfer analysis, but most of the existing numerical simulation methods are based on fixed physical models and parameters, lack of adaptive learning and dynamic deduction ability for complex and variable conditions in actual engineering, and cannot accurately simulate the dynamic evolution process of pile load transfer. SUMMARY

[0004] In view of the above-mentioned problems, in combination with the first aspect of the present application, the embodiments of the present application provide a pile load transfer simulation method combined with machine learning, which comprises: constructing an interaction scenario model between pile foundation and soil, the interaction scenario model comprising pile foundation construction elements, soil hierarchical elements and external load elements; generating a load transfer process fragment set based on the interaction scenario model, the load transfer process fragment set comprising continuous process records of pile foundation and soil interaction under different load conditions; calling a pre-trained load transfer dynamic evolution model to perform inverse learning processing on the load transfer process fragment set, to generate a load transfer path evolution feature sequence and a pile-soil interface state evolution feature sequence; performing load transfer dynamic deduction according to the load transfer path evolution feature sequence and the pile-soil interface state evolution feature sequence, to obtain pile load distribution dynamic characteristics and pile displacement evolution characteristics; generating a pile load transfer simulation analysis report containing load transfer dynamic process curves based on the pile load distribution dynamic characteristics and the pile displacement evolution characteristics.

[0005] In still another aspect, the embodiments of the present application also provide a pile foundation load transfer simulation system combined with machine learning, comprising a processor, a machine readable storage medium, the machine readable storage medium is connected with the processor, the machine readable storage medium is used for storing programs, instructions or codes, and the processor is used for executing the programs, instructions or codes in the machine readable storage medium to realize the above-mentioned method.

[0006] Based on the above aspects, by constructing a detailed interaction scenario model between the pile foundation and the soil mass, key elements such as pile foundation structure, soil hierarchy and external load are comprehensively covered, a load transfer process fragment set is generated based on the interaction scenario model, and a pre-trained load transfer dynamic evolution model is used for inversion learning, which can deeply mine the evolution characteristics of the load transfer path and the pile-soil interface state. The dynamic characteristics of the pile foundation load distribution and the pile displacement are obtained through dynamic deduction, and a simulation analysis report containing dynamic process curves is generated, realizing accurate simulation and comprehensive analysis of the pile foundation load transfer process, thereby improving the accuracy and reliability of the pile foundation load transfer analysis. BRIEF DESCRIPTION OF DRAWINGS

[0007] Figure 1 is the execution flow diagram of the pile foundation load transfer simulation method combined with machine learning provided by the embodiments of the present application.

[0008] Figure 2 is the schematic diagram of exemplary hardware and software components of the pile foundation load transfer simulation system combined with machine learning provided by the embodiments of the present application. DETAILED DESCRIPTION

[0009] The present application will be specifically described below in conjunction with the drawings of the specification, Figure 1 is the flow diagram of the pile foundation load transfer simulation method combined with machine learning provided by an embodiment of the present application, and the pile foundation load transfer simulation method combined with machine learning will be described in detail below.

[0010] Step S110: constructing an interaction scenario model between the pile foundation and the soil mass, the interaction scenario model comprising pile foundation structure elements, soil hierarchy elements and external load elements.

[0011] In this embodiment, the interaction between the concrete precast pile and the surrounding multi-layer soil in a certain building project is taken as an application scenario to construct the interaction scenario model. First, the pile foundation structure elements are determined. The pile foundation is cylindrical, comprising three physical components of pile shaft, pile tip and pile top. The material used for the pile shaft has a specific elastic modulus and Poisson's ratio. The pile tip is a conical structure, and the pile top is provided with a connecting structure to bear external load.

[0012] Then, the soil level elements are determined. The soil in this scenario is divided into first layer soil, second layer soil and third layer soil from the ground surface downwards, each layer of soil has different physical and mechanical properties such as unit weight, internal friction angle and cohesion, and the thickness of each soil layer is different, and there is a clear interface between the soil layers.

[0013] Then, the external load elements are considered. The external load is mainly vertical load, and its action mode includes static load and variable load. The static load is a constant load that acts continuously, and the variable load changes over time according to certain rules, such as periodic change or linear change.

[0014] The above pile foundation structure elements, soil level elements and external load elements are integrated to build a complete interaction scenario model, which can reflect the interaction of the pile foundation under external load in a specific soil environment.

[0015] Step S111: Determine the spatial range boundary of the interaction between the pile foundation and the soil, which includes the spatial region where the pile foundation is located and the spatial region affected by the soil.

[0016] When building the interaction scenario model, the spatial range boundary is first determined. For the spatial region where the pile foundation is located, the center axis of the pile foundation is taken as the reference, and the diameter and length of the pile foundation are considered to determine its coordinate range in three-dimensional space, which should completely contain the entire pile foundation structure.

[0017] For the spatial region affected by the soil, according to the length of the pile foundation and the properties of the surrounding soil, a sufficiently large spatial range is determined, so that the stress and deformation of the soil within this range can be fully considered, and the influence of the soil beyond this range on the pile foundation can be ignored. For example, in the lateral direction, extend to a certain distance on both sides of the center axis of the pile foundation, and in the longitudinal direction, extend to a certain depth below the pile tip from the ground surface. Thus, by determining the above spatial range boundary, a clear region is defined for subsequent model building and simulation calculation.

[0018] Step S112: Divide the pile foundation structure unit and the soil level unit within the spatial range boundary, the pile foundation structure unit corresponds to the physical component of the pile foundation, and the soil level unit corresponds to the physical component of different soil layers.

[0019] Within the determined spatial range boundary, the pile foundation is divided into structure units. The pile body is divided into multiple cylindrical pile foundation structure units along the length direction, each unit has the same length and diameter, and its material properties are consistent with the whole pile body. The pile tip and the pile top are also divided into independent pile foundation structure units to accurately reflect their special structure and stress conditions.

[0020] For the soil layer unit, the soil influence area is divided into soil layer units corresponding to each soil layer according to the division surface of the soil layer. Each soil layer unit has a certain size in the horizontal and vertical directions, and the soil layer units in the same soil layer have the same physical and mechanical properties. For example, the size of the soil layer unit corresponding to the first layer of soil in the horizontal and vertical directions is determined according to the distribution range of the soil layer, and the soil layer units of the second layer of soil and the third layer of soil are divided in a similar manner. Through the above division, the pile foundation and the soil can be analyzed and calculated in the form of units in the subsequent analysis and calculation.

[0021] Step S113: Set the interaction mode rule between the pile foundation construction unit and the soil layer unit, which is used to describe the specific mode of interaction between the pile foundation and the surrounding soil when the pile foundation bears the load.

[0022] When setting the interaction mode rule, the contact relationship between the pile foundation construction unit and the adjacent soil layer unit is considered. When the pile foundation is subjected to a load, the pile body construction unit will generate pressure on the surrounding soil layer unit, and at the same time, the soil layer unit will generate a reaction force on the pile body construction unit. The above interaction is in the form of friction and normal force.

[0023] For the interaction between the pile side pile foundation construction unit and the soil layer unit, the main consideration is the action of friction, and the size of the friction is related to the contact area between the pile body construction unit and the soil layer unit, the internal friction angle of the soil, and the normal pressure of the contact surface. For the interaction between the pile tip construction unit and the corresponding soil layer unit, the main consideration is the action of normal force, and the size of the normal force is related to the load transmitted by the pile tip and the bearing capacity of the soil.

[0024] At the same time, when the pile foundation construction unit is displaced, the soil layer unit will produce a corresponding deformation, which will in turn affect the stress state of the pile foundation construction unit, forming a dynamic interaction relationship.

[0025] Step S114: Configure the initial state parameters of the pile foundation construction unit and the initial state parameters of the soil layer unit, which include the initial stress condition and the initial displacement condition.

[0026] When configuring the initial state parameters of the pile foundation construction unit, for each pile foundation construction unit, the initial stress condition is zero, i.e. when there is no external load acting on it, there is no stress inside the unit. The initial displacement condition is also zero, and each pile foundation construction unit is in its initial position without any displacement.

[0027] For the soil level unit, the initial stress condition considers the self-weight stress of the soil, and the soil level units at different depths have different self-weight stresses. The greater the depth, the greater the self-weight stress, which is related to the specific weight and thickness of the soil above the unit. The initial displacement condition is also zero, and the soil level unit is in a stable position without displacement in the initial state.

[0028] These initial state parameters are input into the interaction scenario model as initial conditions for model simulation.

[0029] Step S115: Validity verification is performed on the interaction scenario model to ensure that the spatial range boundary, the pile foundation construction unit, the soil level unit, the interaction mode rule, and the initial state parameter are mutually adapted.

[0030] When verifying the validity of the interaction scenario model, first check whether the spatial range boundary can completely contain the soil range that may be affected by the pile under the maximum load. If it is found that the boundary is too small to cover the entire affected area, the boundary range needs to be adjusted.

[0031] Next, verify whether the division of the pile foundation construction unit and the soil level unit is reasonable to 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 precision, or too small, increasing the calculation amount, appropriate adjustments are made.

[0032] Then, check whether the interaction mode rule matches the actual physical and mechanical properties of the pile and soil. By comparing the interaction rules in similar existing engineering cases, the rationality of the rule is verified.

[0033] Finally, verify whether the configuration of the initial state parameter is correct to ensure that the initial stress and initial displacement settings conform to the actual situation, such as whether the initial self-weight stress calculation of the soil level unit is accurate.

[0034] Through the above verification steps, it is ensured that the various components of the model are mutually adapted and can accurately simulate the interaction between the pile and the soil.

[0035] Step S120: Based on the interaction scenario model, a load transfer process fragment set is generated, which contains continuous process records of the interaction between the pile and the soil under different load conditions.

[0036] After the construction and validity verification of the interaction scenario model are completed, a load transfer process fragment set is generated based on the model. By setting different load conditions, the model is run for simulation, and the continuous process of the interaction between the pile and the soil is recorded. Then, these process records are processed to form a process fragment set to fully reflect the load transfer characteristics under different load conditions.

[0037] Step S121: setting multiple load action conditions in the interaction scenario model, the load action conditions including different load application positions, different load application manners and different load variation forms.

[0038] Multiple load action conditions are set in the interaction scenario model. The load application positions are mainly the top of the pile, and the case of applying lateral load at different heights of the pile body is also considered.

[0039] The load application manners include concentrated load and uniform load, the concentrated load acting on a specific point at the top of the pile, and the uniform load being uniformly distributed in a certain area at the top of the pile.

[0040] The load variation forms include multiple forms, such as static load, i.e. the load remains constant after being applied; linearly increasing load, the load gradually increases at a certain rate from the initial value to a certain value and then remains stable; and periodically varying load, the load varies according to a certain period and amplitude.

[0041] The above different load application positions, application manners and variation forms are combined to form multiple load action conditions, which are used for simulation of the interaction scenario model.

[0042] Step S122: for each of the load action conditions, running the interaction scenario model for simulation, and recording the entire process information of the pile foundation from the beginning of stress to the stable state, the entire process information including the load transfer of each part of the pile foundation and the stress response of each unit of the soil body.

[0043] For each load action condition, the interaction scenario model is run for simulation. During the simulation process, from the time when the load is applied, every certain time interval, the load transfer of each part of the pile foundation is recorded once, including the load size and direction borne by the pile top, the pile body and the pile tip, and the load transfer path between units.

[0044] At the same time, the stress response of each unit of the soil body is recorded, including the stress size and direction of each soil layer unit and the deformation of the unit.

[0045] The recording continues until the pile foundation and the soil body reach a stable state, i.e. 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 body also tend to be stable, at which time the recording is stopped, and the entire process information under the load action condition is obtained.

[0046] Step S123: segmenting the entire process information according to time sequence for processing, dividing the entire process information into multiple continuous process segments, each process segment corresponding to a specific stage in the load transfer process.

[0047] When all the process information is processed in time sequence, the segmentation points are determined according to the characteristic changes in the load transfer process. For example, in the linearly increasing load stage, when the load increases to a certain specific proportion, it can be taken as a segmentation point; in the constant load stage, when the deformation rate of the pile foundation and the soil body changes obviously, it can also be taken as a segmentation point.

[0048] All the process information from the beginning to the stable state is divided into multiple continuous process segments according to these segmentation points, and 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 stability, the late stage of load stability, etc.

[0049] Each process segment contains the load transfer conditions and the soil force response conditions recorded at all time points in the stage.

[0050] Step S124: identifying key change nodes in each process segment, the key change nodes including nodes where the load transfer direction changes and nodes where the interaction strength between the pile-soil interface changes.

[0051] When identifying the key change nodes in each process segment, for the 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 unit at a certain time changes obviously compared with the previous time, for example, from vertical transfer to both vertical and horizontal transfer, the time is a key change node.

[0052] For the nodes where the interaction strength between the pile-soil interface changes, the friction and normal force between the structural units of the pile body and the hierarchical units of the soil body are monitored, and when the sizes of these forces change suddenly at a certain time, that is, the interaction strength changes, the time is another key change node.

[0053] These key change nodes are marked in the corresponding process segment for subsequent analysis and processing.

[0054] Step S125: classifying and arranging the process segments containing the key change nodes according to the load action conditions to form a load transfer process segment set with time sequence correlation, and each process segment in the load transfer process segment set contains complete load transfer stage characteristics.

[0055] The process segments containing the key change nodes are classified according to the load action conditions, for example, all the process segments belonging to the static load action condition are classified into one category, the process segments of the linearly increasing load action condition are classified into another category, and so on.

[0056] In each category, the process segments are arranged in chronological order so that the end time of the previous process segment is connected with the start time of the next process segment, forming a sequence with chronological correlation.

[0057] The load transfer process segment set formed in this way contains the complete load transfer characteristics of the corresponding stage for each process segment, including load size, transfer direction, pile-soil interaction strength, etc., and is arranged in order according to load conditions and chronological order.

[0058] Step S130: calling the pre-trained load transfer dynamic evolution model to perform inverse learning processing on the load transfer process segment set to generate a load transfer path evolution feature sequence and a pile-soil interface state evolution feature sequence.

[0059] The pre-trained load transfer dynamic evolution model is called, and the generated load transfer process segment set is input into the model. Through inverse learning of the process segment set, the model analyzes the rules of load transfer and the changes of pile-soil interface state, and further generates a load transfer path evolution feature sequence and a pile-soil interface state evolution feature sequence to fully reflect the dynamic process of load transfer.

[0060] Step S131: inputting the load transfer process segment set into the process coding component of the load transfer dynamic evolution model to code and convert the time sequence data in the load transfer process segment set, generating a process coding data sequence, which is used to represent the time sequence features of the load transfer process.

[0061] The load transfer process segment set is input into the process coding component of the load transfer dynamic evolution model. The process coding component first analyzes the time sequence data in each process segment and extracts the characteristic parameters such as load size, transfer direction, and pile-soil interaction strength.

[0062] Then, these characteristic parameters are coded and converted into numerical forms suitable for model processing. For example, the load transfer direction is converted into a corresponding angle value, and the interaction strength is converted into a corresponding dimensionless value.

[0063] The coded characteristic parameters are arranged in chronological order to form the coding data of each process segment, and the coding data of all process segments are connected in chronological order to generate the process coding data sequence of the entire load transfer process, which can completely represent the time sequence features of the load transfer process.

[0064] Step S132: Path feature capturing of the process coding data sequence is performed by a path tracking component of the load transfer dynamic evolution model, and initial characteristics of the load transfer path are generated by analyzing the change of the load transfer path inside the pile foundation and between the pile foundation and the soil.

[0065] The process coding data sequence is processed by the path tracking component of the load transfer dynamic evolution model. The path tracking component analyzes the process coding data sequence segment by segment, and identifies the load transfer path inside the pile foundation from the pile top to the pile body and the pile tip, and the load transfer path from the pile body to the surrounding soil.

[0066] During the analysis, the change of the direction and range of the load transfer path is focused on, such as whether the load transfer is straight or deviates to one side when transferring in the pile body, and whether the transfer range is gradually expanded or kept constant when transferring to the soil.

[0067] According to the analysis results, information representing the path characteristics is extracted, such as the main direction of the path, the range of the covered pile foundation structure unit and soil layer unit, etc., which are combined to form the initial characteristics of the load transfer path.

[0068] Step S1321: A path recognition window is set in the path tracking component, which is used to intercept local time sequence segments in the process coding data sequence.

[0069] The path recognition window is set in the path tracking component, and has a certain time length. For example, the time length of the window can cover several consecutive time points in the process coding data sequence, so that the window can intercept local time sequence segments with a certain time span.

[0070] The size of the window can be set according to the time resolution of the process coding data sequence, so as to ensure that the intercepted local time sequence segments can reflect the basic characteristics of the load transfer path in that time period.

[0071] Step S1322: The path recognition window is used to traverse the process coding data sequence, and the direction of the path is determined for each local time sequence segment to determine the transfer direction and transfer coverage range of the load in the local time sequence segment.

[0072] The path recognition window is used to traverse the process coding data sequence, starting from the starting time point of the sequence, and the window moves backward one time point at a time until the entire sequence is traversed.

[0073] For each intercepted local time series segment, the path tracking component analyzes the relevant data of load transfer therein, determines the transfer direction of the load in the time period, such as vertical downward transfer, horizontal transfer, or oblique transfer, etc.

[0074] Meanwhile, the transfer coverage is determined, i.e. the number and location range of pile foundation construction units and soil body hierarchical units involved in the load transfer, such as which construction units of the pile body and which soil body hierarchical units around are covered.

[0075] Step S1323: The transfer direction and transfer coverage of adjacent local time series segments are compared and analyzed to analyze the continuation and turning situations of the load transfer path, and a path change identifier is generated.

[0076] The transfer direction and transfer coverage of two adjacent local time series segments are compared. If the transfer direction of the latter segment is basically consistent with that of the former segment, and the transfer coverage is basically the same or presents continuous expansion or contraction, it means that the load transfer path is in a continuation situation.

[0077] If the transfer direction of the latter segment is significantly different from that of the former segment, or the transfer coverage presents a non-continuous change, it means that the load transfer path has a turning situation.

[0078] According to the results of the comparison and analysis, a path change identifier is generated, which is marked as a specific identifier for the continuation situation and another specific identifier for the turning situation, to distinguish different path change situations.

[0079] Step S1324: The process coding data sequence is segmented according to the path change identifier to obtain a plurality of path paragraphs with stable transfer paths.

[0080] The process coding data sequence is segmented according to the path change identifier, and when a turning situation path change identifier is encountered, the time point where the identifier is located is taken as a segmentation point.

[0081] In this way, the process coding data sequence is divided into a plurality of continuous parts, each part containing a plurality of continuous local time series segments, and the load transfer path in each part is in a continuation situation, i.e. has a stable transfer direction and transfer coverage, and these parts are path paragraphs with stable transfer paths.

[0082] Step S1325: The transfer direction feature, transfer coverage feature, and duration feature of each path paragraph are extracted and combined to form a load transfer path initial feature, which is used to represent the basic features of the load transfer path in different stages.

[0083] For each path paragraph, its transfer direction feature, i.e. the main direction of load transfer within the paragraph, can be determined by synthesizing the transfer directions of all local time series segments within the paragraph.

[0084] The transfer coverage feature, i.e. the overall range of pile construction units and soil hierarchical units involved in load transfer within the paragraph, is extracted.

[0085] The duration feature, i.e. the length of time experienced by the path paragraph from start to end, is extracted.

[0086] The three features are combined together to form the load transfer path initial feature of the path paragraph, and the initial features of multiple path paragraphs are arranged in time sequence to collectively constitute the load transfer path initial feature of the entire process.

[0087] Step S133: The state association component of the load transfer dynamic evolution model is utilized to perform association analysis processing on the key change nodes in the process coding data sequence, determine the association relationship between the pile-soil interface interaction strength and the load transfer path, and generate an association relationship feature description.

[0088] The state association component of the load transfer dynamic evolution model extracts relevant data of all key change nodes from the process coding data sequence, which includes the pile-soil interface interaction strength and load transfer path features at the key change nodes.

[0089] The state association component analyzes these data to study how the change in the pile-soil interface interaction strength at the key change nodes affects the change in the load transfer path, and how the change in the load transfer path affects the change in the pile-soil interface interaction strength. For example, when the pile-soil interface interaction strength increases, analyze whether the load transfer path will become more concentrated or the transfer direction will be deflected; when the load transfer path expands to a new soil hierarchical unit, analyze whether the pile-soil interface interaction strength will change accordingly. Through the above two-way analysis, the association pattern between the two is determined, and then an association relationship feature description that can quantify the above association relationship is generated.

[0090] Step S1331: Extract the state description data of all key change nodes from the process coding data sequence, which includes the pile-soil interface interaction strength parameter and the load transfer path parameter.

[0091] When extracting the state description data of all key change nodes from the process coding data sequence, first locate the position of each key change node in the sequence. For each key change node, extract its corresponding pile-soil interface interaction strength parameter, which includes the size of pile side friction, the size of pile tip normal force, etc., which can reflect the interaction strength of the pile-soil interface at this node.

[0092] At the same time, extract the load transfer path parameter at the key change node, which includes the main direction angle of load transfer, the range information of the pile foundation construction unit and soil hierarchical unit covered by the transfer, etc., which can describe the characteristics of the load transfer path at this node.

[0093] Combine the extracted pile-soil interface interaction strength parameter and load transfer path parameter to form the state description data of each key change node.

[0094] Step S1332: Perform standardization conversion processing on the state description data, input the standardized state data into the correlation analysis model of the state correlation 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.

[0095] Perform standardization conversion processing on the state description data, convert the pile-soil interface interaction strength parameter and the load transfer path parameter into the same numerical interval, and eliminate the influence between different parameters due to the difference in dimension and numerical range. For example, convert the parameter value to a value within a certain interval according to the set rule.

[0096] Input the standardized state data into the correlation analysis model of the state correlation component, which will perform pairwise analysis on the pile-soil interface interaction strength parameter and the load transfer path parameter. Calculate the mutual information value between the two, which is used to measure the dependence between the two parameters, the larger the value, the stronger the dependence; calculate the correlation coefficient value between the two, which is used to measure the linear correlation degree between the two parameters, the value range is within a certain interval, positive number indicates positive correlation, negative number indicates negative correlation, and the larger the absolute value, the higher the correlation degree.

[0097] Step S1333: Determine the correlation tightness between parameters according to the mutual information value and the correlation coefficient value, and screen out parameter combinations with correlation tightness exceeding a preset threshold.

[0098] The correlation closeness between parameters is determined according to the calculated mutual information value and correlation coefficient value. The mutual information value and the correlation coefficient value are combined to comprehensively judge, for example, when the mutual information value is greater than a preset value and the absolute value of the correlation coefficient value is greater than a preset value, it is determined that the corresponding parameter combination has a relatively high correlation closeness.

[0099] A preset threshold is set, which is determined according to engineering experience and model training requirements. Parameter combinations with a correlation closeness exceeding the preset threshold are screened out, and these parameter combinations can reflect the important correlation relationship between the pile-soil interface interaction strength and the load transfer path.

[0100] Step S1334: An association relationship network structure is constructed according to the screened parameter combinations, wherein a node in the association relationship network structure represents a parameter, an edge represents the association relationship between parameters, and the weight of the edge represents the correlation closeness.

[0101] An association relationship network structure is constructed according to the screened parameter combinations, and each node in the network structure represents a parameter, including both the pile-soil interface interaction strength parameter and the load transfer path parameter.

[0102] For each screened parameter combination, an edge is established between the corresponding two nodes to represent the association relationship between the two parameters. The weight of the edge is determined according to the mutual information value and the correlation coefficient value of the parameter combination, and the higher the correlation closeness, the greater the weight value of the edge.

[0103] In the above manner, the association relationship between parameters is intuitively presented in the form of a network structure, forming a complete association relationship network structure.

[0104] Step S1335: The association relationship network structure is converted into an association relationship feature description containing node attributes and edge attributes, and the association relationship feature description is used to quantitatively represent the association relationship between the pile-soil interface interaction strength and the load transfer path.

[0105] When converting the association relationship network structure into the association relationship feature description, first, the attributes of each node in the network structure are extracted, including the parameter name represented by the node and the physical meaning of the parameter.

[0106] Then, the attributes of each edge are extracted, including the names of the two nodes connected by the edge, the weight value of the edge, and the mutual information value and the correlation coefficient value corresponding to the weight value.

[0107] The node attributes and edge attributes are arranged and recorded in a set format to form a correlation relationship feature description. The correlation relationship feature description can clearly quantitatively represent the correlation relationship between the pile-soil interface interaction strength and the load transfer path, including information such as which parameters are correlated and how close the correlation is.

[0108] Step S134: dynamically adjusting the initial characteristics of the load transfer path according to the correlation relationship feature description to obtain a load transfer path evolution characteristic sequence that can reflect the influence of pile-soil interaction.

[0109] The initial characteristics of the load transfer path are dynamically adjusted according to the correlation relationship feature description. For example, when the correlation relationship feature description indicates that a load transfer path parameter is strongly correlated with a pile-soil interface interaction strength parameter, the path characteristics corresponding to the load transfer path parameter are adjusted according to the changes in the pile-soil interface interaction strength.

[0110] If the pile-soil interface interaction strength increases and the correlation relationship shows that this will lead to more concentrated load transfer paths, the description of the transfer range in the initial characteristics of the load transfer path is adjusted to make the range smaller and more concentrated.

[0111] Through the above dynamic adjustment, the load transfer path characteristics can reflect the influence of pile-soil interaction. The adjusted path characteristics are arranged in chronological order to obtain a load transfer path evolution characteristic sequence.

[0112] Step S135: extracting the state characteristics of the pile-soil interface at different stages according to the load transfer path evolution characteristic sequence, and arranging them in chronological order to generate a pile-soil interface state evolution characteristic sequence, which contains the interface action type and action strength characteristics of each stage.

[0113] According to the load transfer path evolution characteristic sequence, the action of the load transfer path and the pile-soil interface at different time stages is analyzed. At each stage, the state characteristics of the pile-soil interface are extracted, including the interface action type, such as whether the main action is friction or normal force, or the combined action of both.

[0114] At the same time, the interface action strength characteristics are extracted, including the average size and variation amplitude of the friction and normal force in this stage. These state characteristics are arranged in chronological order to generate a pile-soil interface state evolution characteristic sequence that fully presents the state changes of the pile-soil interface at different stages during the entire load transfer process.

[0115] Step S140: performing dynamic deduction of load transfer according to the load transfer path evolution characteristic sequence and the pile-soil interface state evolution characteristic sequence, to obtain pile foundation load distribution dynamic characteristics and pile body displacement evolution characteristics.

[0116] Dynamic deduction of load transfer is performed using the load transfer path evolution characteristic sequence and the pile-soil interface state evolution characteristic sequence. During the deduction process, the information provided by the load transfer path evolution characteristic sequence and the pile-soil interface state evolution characteristic sequence is combined to simulate the transfer of load at different time stages and the response of the pile foundation and soil, and then the pile foundation load distribution dynamic characteristics reflecting the change of pile foundation load distribution over time and the pile body displacement evolution characteristics reflecting the change of pile body displacement over time are obtained.

[0117] Step S141: inputting the load transfer path evolution characteristic sequence and the pile-soil interface state evolution characteristic sequence into a dynamic deduction module, initializing deduction environment parameters, setting deduction time interval and deduction total duration.

[0118] The load transfer path evolution characteristic sequence and the pile-soil interface state evolution characteristic sequence are input into the dynamic deduction module. The dynamic deduction module first initializes the deduction environment parameters, including setting the physical and mechanical property parameters of the pile foundation construction units and the soil layer units, such as the elastic modulus of the pile foundation material and the specific gravity of the soil, which are consistent with the parameters in the interaction scenario model.

[0119] The deduction time interval, i.e. the time step for each deduction calculation, is set. This interval should be small enough to ensure accurate capture of subtle changes in the load transfer process. The deduction total duration is set, which should cover the entire process from the start of load action to the stabilization of the pile foundation and soil.

[0120] Step S142: in each deduction time interval, determine the transfer direction and transfer proportion of the load in the pile foundation and soil according to the load transfer path evolution characteristic sequence at the current time.

[0121] In each deduction time interval, the dynamic deduction module reads the information in the load transfer path evolution characteristic sequence at the current time. According to these information, the direction of load transfer from one construction unit to another construction unit inside the pile foundation, and the direction of load transfer from the pile foundation construction unit to the adjacent soil layer unit are determined.

[0122] At the same time, the transfer proportion of the load on different transfer paths is determined, for example, what proportion of the total load is transferred along the pile body to the pile tip, and what proportion is transferred through the pile side to the surrounding soil layer units.

[0123] Step S143: determining the action type and action strength of the pile-soil interface based on the current time's pile-soil interface state evolution characteristic sequence, and adjusting the load transfer efficiency according to the action type and the action strength.

[0124] Based on the current time's pile-soil interface state evolution characteristic sequence, the action type of the pile-soil interface is determined, such as whether the pile side friction or the pile tip normal force is the main action at this time. At the same time, the action strength, i.e. the size of the friction and the normal force, is determined. According to the action type and the action strength, the load transfer efficiency is adjusted, for example, when the action strength is large, the load transfer efficiency is high, that is, more load can be transmitted through the interface; when the action strength is small, the load transfer efficiency is low.

[0125] Step S144: calculating the load increase of each position of the pile foundation and the stress increase of each unit of the soil body according to the transfer direction, the transfer proportion and the load transfer efficiency.

[0126] According to the transfer direction, the flow path of the load in the pile foundation and the soil body is determined; according to the transfer proportion, the total load in the time interval is distributed to each transfer path; and combined with the load transfer efficiency, the actual effective load transfer amount on each transfer path is calculated.

[0127] For each position of the pile foundation, the difference between the load amount received in the time interval and the load amount transferred out is calculated to obtain the load increase; for each unit of the soil body, the effective load transfer amount transferred to the unit is combined with the property parameters of the soil body to calculate the stress increase.

[0128] Step S1441: determining the distribution path of the load between the pile foundation construction unit and the soil body hierarchical unit based on the transfer direction, and generating a load distribution path graph.

[0129] Based on the transfer direction, the distribution path of the load between the pile foundation construction unit and the soil body hierarchical unit is drawn to form a load distribution path graph. In the graph, the transfer direction of the load is represented by a line, the starting point of the line is the unit where the load flows out, and the ending point of the line is the unit where the load flows in. The distribution path of the load is intuitively displayed through the above graph.

[0130] Step S1442: distributing the total load to each branch path in the load distribution path graph according to the transfer proportion to obtain the load distribution value of each branch path.

[0131] According to the transfer proportion, the total load value in the deduced time interval is distributed to each branch path in the load distribution path graph. For example, if the transfer proportion of a branch path is a certain proportion, the load distribution value of the branch path is equal to the total load value multiplied by the proportion. In this way, the load distribution values of all branch paths are obtained.

[0132] Step S1443: determining the load transfer efficiency coefficient of each branch path according to the action intensity parameter in the pile-soil interface state evolution characteristic sequence, the load transfer efficiency coefficient being positively correlated with the action intensity parameter.

[0133] The load transfer efficiency coefficient of each branch path is determined according to the action intensity parameter in the pile-soil interface state evolution characteristic sequence, such as the size of friction and normal force. The greater the action intensity parameter, the greater the corresponding load transfer efficiency coefficient, and the two are positively correlated. For example, when the action intensity parameter is a certain level, the load transfer efficiency coefficient is a certain corresponding value.

[0134] Step S1444: multiplying 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.

[0135] The load distribution value of each branch path is multiplied by the corresponding load transfer efficiency coefficient to obtain the effective load transfer value of the branch path in the time interval, which represents the actual load amount that can be transferred through the path.

[0136] Step S1445: calculating the load increase of each structural unit of the pile foundation according to the effective load transfer value, the load increase being equal to the effective load transfer value flowing into the structural unit minus the effective load transfer value flowing out of the structural unit.

[0137] For each structural unit of the pile foundation, all the effective load transfer values flowing into the unit are summed up, and then all the effective load transfer values flowing out of the unit are subtracted to obtain the load increase of the structural unit in the time interval.

[0138] Step S1446: calculating the stress increase of each unit of the soil based on the effective load transfer value and the stiffness parameter of the soil hierarchical unit, the stress increase being positively correlated with the effective load transfer value and negatively correlated with the stiffness parameter.

[0139] The stress increase is calculated based on the effective load transfer value and the stiffness parameter of the soil hierarchical unit. The greater the effective load transfer value, the greater the stress increase; the greater the stiffness parameter of the soil hierarchical unit, the smaller the stress increase. Through the mutual relationship between the two, the stress increase of each unit of the soil in the time interval is calculated.

[0140] Step S145: superimposing the load increase and the stress increase on the load value and the stress value at the previous moment to obtain the load distribution state and the stress distribution state at the current moment.

[0141] The load increase amount of each position of the pile foundation calculated is added to the load value of the position at the previous time to obtain the load value of each position of the pile foundation at the current time, and the load distribution state at the current time is formed by combination; the stress increase amount of each unit of the soil body is added to the stress value of the unit at the previous time to obtain the stress value of each unit of the soil body at the current time, and the stress distribution state at the current time is formed by combination.

[0142] Step S146: record the load distribution state and the stress distribution state of each deduction time interval in time sequence, extract the characteristics of the load distribution state and the stress distribution state changing with time, generate the pile foundation load distribution dynamic characteristics and the pile body displacement evolution characteristics, and the pile foundation load distribution dynamic characteristics and the pile body displacement evolution characteristics contain load evolution trend and displacement evolution trend information.

[0143] In time sequence, the load distribution state and the stress distribution state at the end of each deduction time interval are recorded in turn. The changes of these states with time are analyzed, and the change characteristics are extracted, such as the increase and decrease trend of the load at different positions of the pile foundation, the diffusion trend of the stress in different units of the soil body, etc.

[0144] According to the change characteristics of the load distribution state, the pile foundation load distribution dynamic characteristics are generated, and according to the stress distribution state, the displacement change of the pile body is analyzed in combination with the deformation characteristics of the soil body and the pile foundation, the displacement change trend is extracted, and the pile body displacement evolution characteristics are generated.

[0145] Step S150: generate a pile foundation load transfer simulation analysis report containing a load transfer dynamic process curve based on the pile foundation load distribution dynamic characteristics and the pile body displacement evolution characteristics.

[0146] The pile foundation load distribution dynamic characteristics and the pile body displacement evolution characteristics are combined, and these characteristics are presented in a suitable form, including drawing the load transfer dynamic process curve, and then combining with the text description and analysis to generate a complete pile foundation load transfer simulation analysis report.

[0147] Step S151: extract the evolution information of the load transfer path from the pile foundation load distribution dynamic characteristics and the pile body displacement evolution characteristics, and the evolution information contains the transfer direction, transfer coverage range and transfer intensity information at different time points.

[0148] From the pile foundation load distribution dynamic characteristics, the distribution of the load in the pile foundation at different time points is analyzed, and the transfer direction and the transfer coverage range of the load are inferred; in combination with the pile body displacement evolution characteristics, the accuracy of the transfer direction is further verified.

[0149] At the same time, according to the density and size of the load distribution, the transfer intensity information at different time points is determined. These transfer direction, transfer coverage range and transfer intensity information are summarized to form the evolution information of the load transfer path.

[0150] For example, step S1511: data analysis processing is performed on the pile load distribution dynamic characteristics and the pile body displacement evolution characteristics to separate out characteristic components related to the load transfer path.

[0151] The data analysis processing is performed on the pile load distribution dynamic characteristics and the pile body displacement evolution characteristics to remove irrelevant interference information such as some local and short-term load fluctuations or displacement fluctuations.

[0152] The characteristic components directly related to the load transfer path are retained and extracted, such as the distribution change of the load in the main transfer direction and the displacement change of the pile body corresponding to the transfer path.

[0153] Step S1512: transfer direction parameters at all time points are extracted from the characteristic components, and the transfer direction parameters are used to represent the transfer angle of the load in space.

[0154] From the separated characteristic components, the main direction of load transfer is determined for each time point, and a transfer direction parameter is used to represent it, which is an angle value that can accurately reflect the transfer angle of the load in space.

[0155] Step S1513: transfer coverage range parameters at all time points are extracted, and the transfer coverage range parameters are used to represent the length range of the pile foundation and the depth range of the soil covered by the load transfer.

[0156] For each time point, the transfer coverage range parameter is extracted, which includes the length range of the pile foundation involved in the load transfer, i.e., the length interval from the pile top to a certain position of the pile body, and the depth range of the soil involved, i.e., the interval from the ground surface to a certain depth of the soil, which clearly represents the transfer coverage range through these parameters.

[0157] Step S1514: transfer intensity parameters at all time points are extracted, and the transfer intensity parameters are used to represent the size of the load transfer.

[0158] At each time point, the transfer intensity parameter is extracted according to the load distribution dynamic characteristics, which can be the average load size on the transfer path at that time point, or the maximum load value, etc., used to represent the size of the load transfer.

[0159] Step S1515: the transfer direction parameters, the transfer coverage range parameters, and the transfer intensity parameters are paired according to the time points to form evolution data records containing time markers, transfer direction parameters, transfer coverage range parameters, and transfer intensity parameters.

[0160] The transfer direction parameter, the transfer coverage parameter and the transfer intensity parameter at each time point are paired with the corresponding time mark to form an evolution data record, and each record completely contains the characteristic parameters of the load transfer path at the time point.

[0161] Step S1516: The evolution data records are sorted according to the chronological order of the time marks to generate a complete evolution data set, and the evolution data set is used to draw a load transfer dynamic process curve.

[0162] All the evolution data records are sorted according to the chronological order of the time marks to form a complete evolution data set. The evolution data set clearly presents the characteristics of the load transfer path at each time point in chronological order.

[0163] Step S152: The evolution information is input into a curve generation component to draw a dynamic process curve of the load transfer path over time, the horizontal axis of the dynamic process curve represents time, the vertical axis represents transfer intensity, and the shape of the curve represents the changes of the transfer direction and the transfer coverage.

[0164] The evolution information is input into a curve generation component, and the component draws a basic curve framework according to the time marks and the transfer intensity parameters in the evolution data set, with the horizontal axis representing time and the vertical axis representing transfer intensity.

[0165] At the same time, the changes of the transfer direction and the transfer coverage are represented by the shape of the curve, for example, the inclination of the curve represents the change of the transfer direction, and the width of the curve represents the size of the transfer coverage, and the greater the width, the wider the coverage.

[0166] Step S153: The change information of the pile-soil interface action state is extracted from the pile foundation load distribution dynamic characteristics and the pile shaft displacement evolution characteristics, and a state change time sequence table is generated by arranging the time sequence, which contains the action type, action intensity and corresponding load transfer characteristics of each time interval.

[0167] The action type of the pile-soil interface in different time intervals is extracted from the pile foundation load distribution dynamic characteristics, for example, in a certain time interval, the pile-soil interface mainly exhibits friction force, while in another time interval, friction force and normal force may exist simultaneously.

[0168] At the same time, the action intensity in each time interval is extracted, which is embodied by the size of the force transferred by the pile-soil interface, and can be determined according to the interaction force data between the pile body construction unit and the soil layer unit in the pile foundation load distribution dynamic characteristics. In addition, combined with the pile shaft displacement evolution characteristics, the corresponding load transfer characteristics in each time interval are determined, such as the main direction of load transfer and the range of transfer.

[0169] The extracted action types, action intensities, and corresponding load transfer characteristics are arranged in time sequence to generate a state change time sequence table. Each row in the table corresponds to the relevant information of a time interval, making the changes in the pile-soil interface action state clear.

[0170] Step S154: Visualize the dynamic process curve and the state change time sequence table, add titles, coordinate axis labels, and legend explanations, and generate a visualization chart.

[0171] When visualizing the dynamic process curve, add a suitable title to the curve. The title accurately reflects the content represented by the curve, such as "Pile Foundation Load Transfer Path Dynamic Process Curve with Time Change".

[0172] Add the label "Time" to the horizontal axis and indicate the unit of time, such as "seconds". Add the label "Transfer Intensity" to the vertical axis and indicate the corresponding dimensionless unit.

[0173] For the transfer direction and transfer coverage range represented by the curve shape, add a legend explanation, such as using different curve inclination angles to identify the direction and using curve width to identify the coverage range.

[0174] When visualizing the state change time sequence table, add the title "Pile-soil Interface Action State Change Time Sequence Table" to the table, and clearly label the column titles in the table, such as "Time Interval", "Action Type", "Action Intensity", "Load Transfer Characteristics", etc.

[0175] Through these visualizations, the dynamic process curve and the state change time sequence table are more clear and easy to understand, forming a complete visualization chart.

[0176] Step S155: Integrate the visualization chart with the textual description information of the pile foundation load distribution dynamic characteristics and the pile displacement evolution characteristics, and arrange according to the preset report format to generate a pile foundation load transfer simulation analysis report containing the load transfer dynamic process curve.

[0177] Collect the textual description information of the pile foundation load distribution dynamic characteristics, which includes the law of load distribution on each structural unit of the pile foundation with time, the position and time of the maximum load, etc.

[0178] Collect the textual description information of the pile displacement evolution characteristics, including the displacement trend of each part of the pile with time, the position and time of the maximum displacement value, etc.

[0179] Integrate these textual description information with the previously generated visualization charts, and typeset according to the preset report format, which usually includes the following parts: abstract, introduction, simulation model introduction, load transfer process analysis, result visualization, conclusion, etc.

[0180] During the typesetting process, dynamic process curves and state change timing tables are inserted into the corresponding analysis sections, and the textual description information is explained in detail around the charts, so that readers can fully understand the simulation results of pile load transfer through the combination of text and charts.

[0181] Finally, a complete pile load transfer simulation analysis report is generated, which contains load transfer dynamic process curves.

[0182] Further, the embodiment can also include the step of training the load transfer dynamic evolution model.

[0183] In order to enable the load transfer dynamic evolution model to accurately process the load transfer process fragment set and generate the corresponding evolution feature sequence, the load transfer dynamic evolution model needs to be trained, and the specific steps are as follows.

[0184] Step S211: Collect a large amount of sample data of pile-soil interaction, which contains load transfer process records under different pile types, different soil environments and different load conditions.

[0185] Through consulting engineering archives, laboratory tests and numerical simulation, a large amount of sample data is collected. The pile types in the sample data include piles of different materials, sizes and structures, such as reinforced concrete piles, wood piles, steel pipe piles, etc.; different soil environments cover various soil layer combinations and large differences in soil physical and mechanical properties; different load conditions include various load sizes, application methods and change forms.

[0186] Each sample data contains complete load transfer process records, such as the load transfer of each part of the pile and the stress response of each unit of the soil, as described in the foregoing, to ensure that the sample data is diverse and representative, and can cover a variety of possible engineering scenarios.

[0187] Step S212: Preprocess the collected sample data, including data cleaning, data standardization and data division.

[0188] Check each record in the sample data one by one, identify noise data, which may be abnormal fluctuation values caused by measurement device errors or data recording errors, and remove noise through smoothing processing, etc.

[0189] For outliers that deviate significantly from the normal range, such as load values at certain time points that are much larger than those at other time points and do not conform to the load variation law, they are excluded.

[0190] For data records with missing values, interpolation is used to fill in the missing values according to the trend before and after the data, ensuring the integrity and continuity of the sample data.

[0191] The sample data contains various characteristic parameters, such as load size (unit: force), displacement (unit: length), stress (unit: pressure), etc., which have different dimensions.

[0192] Standardization is used to process the above characteristic parameters, converting each characteristic parameter into a dimensionless value, so that different characteristic parameters are comparable. For example, for load size, it is converted to a proportion relative to the maximum load value in the sample data; for displacement, it is converted to a proportion relative to the length of the pile foundation, etc.

[0193] Then, according to the preset proportion, such as seventy percent, fifteen percent, fifteen percent, the standardized sample data is randomly divided into training set, validation set and test set. The training set is used for parameter learning and training of the model; the validation set is used to evaluate the performance of the model during training, and adjust the hyperparameters of the model, such as learning rate, network layer number, etc.; the test set is used to evaluate the generalization ability of the model after training, and test the performance of the model on unseen data.

[0194] Step S213: Construct the network structure of the load transfer dynamic evolution model, including process coding components, path tracking components, state association components, etc. and set the initial parameters of each module.

[0195] Construct the process coding component, use the recurrent neural network structure, which can effectively process time series data, set the number of hidden layers, the number of neurons in each hidden layer, etc. initial parameters, so that it can effectively encode the time series data of the load transfer process.

[0196] Construct the path tracking component, use the convolutional neural network structure, set different sizes of convolution kernels to extract local features in the process coding data sequence, so as to capture the features of the load transfer path, set the number of convolution layers, the size of the convolution kernel, etc. initial parameters.

[0197] Construct the state association component, use the graph neural network structure, which is suitable for processing data with association relationship, set the feature dimension of the graph node, the connection mode of the edge, etc. initial parameters, to realize the analysis of the association relationship between the pile-soil interface interaction strength and the load transfer path.

[0198] At the same time, the initial parameters of the model, such as the optimizer type, loss function, etc. 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.

[0199] Step S214: Input the training set into the constructed load transfer dynamic evolution model for training, and continuously adjust the model parameters through the back propagation algorithm.

[0200] The sample data in the training set is input into the load transfer dynamic evolution model in batches, and the model processes the input data to generate the prediction results of the load transfer path evolution feature sequence and the pile-soil interface state evolution feature sequence.

[0201] Compare the prediction results with the actual results in the sample data, calculate the value of the loss function, and adjust the parameters of each module from the output layer to the input layer through the back propagation algorithm, such as the weights of the recurrent neural network and the convolution kernel parameters of the convolutional neural network.

[0202] Repeat the above process and continuously iterate the training until the value of the loss function reaches the preset threshold or the training reaches the maximum number of iterations.

[0203] Step S215: Use the validation set to evaluate the model during training, and adjust the hyperparameters of the model according to the evaluation results.

[0204] After each iteration period of model training, input the validation set into the current model to obtain the validation results.

[0205] Evaluate the performance of the model on the validation set by calculating the error indicators between the validation results and the actual results of the validation set, such as mean absolute error and mean square error.

[0206] According to the evaluation results, adjust the hyperparameters of the model, such as increasing or decreasing the number of neurons in the hidden layer, adjusting the size of the learning rate, etc., so that the performance of the model on the validation set is optimal.

[0207] Step S216: When the model training is completed, use the test set to perform the final evaluation of the model, test the generalization ability of the model, and if the evaluation results meet the preset requirements, the model training is completed.

[0208] Input the test set into the trained model to obtain the test results, and calculate the error indicators between the test results and the actual results of the test set.

[0209] If the error indicators are within the preset acceptable range, it means that the model has good generalization ability and can accurately analyze and predict the load transfer process that has not been seen before, and the model training is completed.

[0210] If the error index exceeds the preset range, the sample data needs to be rechecked or the model structure needs to be adjusted, and the training and evaluation need to be performed again until the model meets the preset requirements.

[0211] To ensure the accuracy and reliability of the pile load transfer simulation analysis report, it needs to be verified, and the specific steps are as follows.

[0212] 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 under the action of actual load.

[0213] Through research, select an actual engineering case similar to the type of pile foundation, soil environment and load condition in this embodiment, and the actual engineering case should have complete monitoring data.

[0214] Collect the measured data obtained by monitoring equipment during the construction and use of the engineering case, such as load change data of pile top, stress monitoring data at different depths of pile body, displacement monitoring data of pile body, and stress and deformation data of surrounding soil, etc.

[0215] Ensure that the collected measured data has sufficient time span and data accuracy, which can reflect the actual load transfer process and the response of pile foundation and soil.

[0216] 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 them.

[0217] Extract key prediction results from the simulation analysis report, such as load distribution state of pile foundation at different time points, displacement evolution characteristics of pile body, etc.

[0218] Compare these prediction results with the measured data of the actual engineering case at the same time point and position, calculate the deviation value of each comparison point, and the deviation value is the difference between the prediction result and the measured data.

[0219] Statistical deviation values of all comparison points, calculate average deviation, maximum deviation and other statistical indicators, and comprehensively understand the difference between simulation results and measured data.

[0220] Step S313: According to the deviation analysis result, evaluate the accuracy of the simulation analysis report, if the deviation is within the acceptable range, determine that the simulation analysis report is effective; if the deviation exceeds the acceptable range, return to the corresponding step for correction.

[0221] According to the requirements of engineering practice for the accuracy of simulation results, set the acceptable range of deviation.

[0222] The calculated average deviation, maximum deviation and other statistical indicators are compared 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 load transfer condition, and the report is effective.

[0223] If there are indicators that exceed the acceptable range, analyze the causes of the deviation. It may be that the parameter setting of the interaction scenario model is unreasonable, the training of the load transfer dynamic evolution model is insufficient, or the parameter adjustment in the dynamic deduction process is improper, etc.

[0224] According to the cause of the deviation, return to the corresponding step for correction, such as re-adjusting the soil parameters of the interaction scenario model, re-training the load transfer dynamic evolution model or optimizing the parameters of the dynamic deduction, etc. After the correction is completed, the simulation analysis report is regenerated and verified again until the report meets the accuracy requirements.

[0225] In this way, the pile load transfer simulation analysis report that passes the verification can be stored and archived, and the index information of the report can be established for subsequent query and call.

[0226] For example, after the simulation analysis report passes the verification, it needs to be stored and archived. First, a unique identifier is assigned to the report, which contains the date of report generation, the corresponding load action condition type and the model version, etc. to distinguish different reports.

[0227] Next, the electronic document of the report is stored in the designated database or file server, and the storage path is set according to the set classification rules, such as classification storage according to the project name, analysis time, etc. At the same time, the index information of the report is established, including the identifier of the report, the generation time, the summary of the load action condition, the main conclusions, etc. These index information will be stored in the index database, which makes it easy for users to query and quickly locate the required report through keywords.

[0228] In addition, the stored report is backed up regularly to prevent data loss. The frequency of backup can be determined according to the importance and update frequency of the report. The backup files are also managed according to the same classification rules and index information to ensure that the report data can be recovered in time when needed.

[0229] Among them, the stored pile foundation load transfer simulation analysis reports can also be reviewed and summarized regularly to extract common rules and special cases. For example, regularly review and summarize the stored simulation analysis reports, such as every certain time, all reports generated in this period are collected and combed. In the review process, compare the pile foundation load transfer characteristics under different load conditions, analyze the common rules, such as whether the evolution trend of the pile foundation load transfer path is similar in the same type of soil, whether the change of the pile-soil interface action state follows certain rules, etc.

[0230] At the same time, pay attention to special cases, that is, those cases whose load transfer characteristics are significantly different from the regular situation, analyze the reasons for the difference, such as special soil distribution, complex load change form, etc.

[0231] The extracted common rules and special cases are arranged into report summary documents, which record the specific performance of the rules, the analysis of the special cases, and the corresponding conclusions. These summary documents will serve as reference materials for subsequent pile foundation design and analysis, helping designers to refer to previous simulation analysis results when designing similar projects, optimize the design scheme, and improve the rationality and reliability of the design.

[0232] In addition, the load transfer dynamic evolution model needs to be maintained and updated regularly to adapt to different engineering scenarios and new technical requirements.

[0233] In order 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, set the maintenance and update cycle, which can be determined according to the frequency of model use, the speed of change of engineering scenarios, and the situation of technological development.

[0234] During the maintenance process, check whether each component of the model is running normally, such as whether the functions of process coding components, path tracking components, state association components, etc. are intact, and whether the accuracy of data processing meets the requirements. If it is found that the components have operation abnormalities or data processing errors, timely repair and debugging, such as correcting algorithm logic errors in components, optimizing data processing procedures, etc.

[0235] For the update of the model, the following aspects are mainly included: first, according to the new engineering practice data and research results, the parameter settings of the model are adjusted, such as the size of the path identification window, the threshold in the correlation analysis, etc., in order to improve the adaptability of the model to new scenarios; second, new feature parameters are introduced or existing feature extraction methods are improved, for example, when some new soil parameters are found to have important influence on load transfer, they are included in the analysis range of the model, and the processing logic of the process coding component and the state correlation component is adjusted accordingly; third, the structure of the model is optimized, such as increasing the number of layers of the neural network or adjusting the number of neurons in each layer, in order to improve the analysis accuracy and the ability to process complex data of the model.

[0236] After completing the maintenance and update of the model, the updated model needs to be tested, and the performance of the model is verified using new test data sets to see if it meets the requirements, and the test content includes the analysis accuracy and running efficiency of the model. Only the model that passes the test can be put into use, and the updated model version information is recorded in the model management document, and the simulation analysis report generated based on the old version of the model is marked to facilitate differentiation.

[0237] Further, new pile foundation engineering case data and load transfer test data can be collected as new training samples to optimize and improve the load transfer dynamic evolution model.

[0238] New pile foundation engineering case data and load transfer test data are continuously collected, which include the construction parameters of pile foundation, the physical and mechanical property parameters of soil, the applied load conditions, the stress and deformation monitoring data of pile foundation and soil, and the detailed data obtained from pile foundation load transfer tests in the laboratory, such as stress distribution of pile body at different load stages and friction force change at pile-soil interface.

[0239] The collected data is preprocessed, first, the abnormal values and error data in the data are removed, for example, by comparing the reasonable range of the same type of data, the obviously deviated data are identified and corrected or deleted. Then, the data is standardized, the parameters of different dimensions are converted into dimensionless parameters, for example, the load size, soil density and other parameters are scaled according to the set proportion to make them in the same numerical range, in order to meet the training requirements of the model.

[0240] The preprocessed new data is used as new training samples, which are merged with the original training samples to form an updated training data set. In the merging process, the format and feature parameters of the new samples and the original samples are ensured to be consistent, in order to facilitate the unified training of the model. At the same time, the training data set is divided, part of it is used as training set for parameter optimization of the model, and the other part is used as validation set for evaluation of the improvement effect of the model.

[0241] According to the new training samples, the load transfer dynamic evolution model is retrained in an incremental training manner to improve the generalization ability and analysis accuracy of the model.

[0242] The load transfer dynamic evolution model is retrained in an incremental training manner, that is, based on the trained model parameters, the model is further trained using new training samples, rather than retraining the entire model. First, load the parameters of the currently used model as the initial parameters for retraining.

[0243] Then, set the parameters for incremental training, such as learning rate, training iteration number, etc. The learning rate can be slightly lower than that during initial training to avoid excessive interference with learned knowledge. The training iteration number is determined according to the number of new training samples and the convergence of the model.

[0244] During training, new training samples are input into the model, and the model adjusts the weight parameters of each layer in a backpropagation manner, so that the model can gradually adapt to the information contained in the new samples. At the same time, during each iteration, the performance of the model is evaluated using the validation set, and the prediction error of the model is calculated. When the prediction error reaches the preset threshold or the training iteration number reaches the set value, the training is stopped.

[0245] After training is completed, the updated model parameters are saved, and the performance of the model is tested. The analysis results of the model before and after updating are compared to evaluate whether the generalization ability and analysis accuracy of the model have been improved. If the model performance meets the requirements, it is used as the new version; if it does not meet the requirements, the parameters for incremental training are adjusted or the number of training samples is increased, and the training is restarted.

[0246] A model performance evaluation index system is established to evaluate the performance of the load transfer dynamic evolution model regularly to ensure the effectiveness and reliability of the load transfer dynamic evolution model.

[0247] A model performance evaluation index system is established, which includes multiple evaluation indexes, such as prediction error rate, i.e. the error proportion between the model's predicted pile load distribution and displacement evolution results and actual monitoring data or test data; model stability, i.e. the consistency degree of the output results when the model runs the same input data multiple times; calculation efficiency, i.e. the time required for the model to process a certain amount of data; generalization ability, i.e. the analysis accuracy of the model on new data not involved in training, etc.

[0248] For each evaluation index, set corresponding evaluation standards and thresholds, such as acceptable threshold of prediction error rate, minimum requirement of model stability, etc. These standards and thresholds are determined according to the needs of engineering practice and the application scenarios of the model.

[0249] The performance of the load transfer dynamic evolution model is evaluated periodically according to the index system, and the evaluation period is consistent with the maintenance and update period of the model. During the evaluation process, a representative test data set is selected, input into the model, and the output results of the model are obtained. Then, according to the calculation method of each evaluation index, the index values of the model on the test data set are calculated.

[0250] The calculated index values are compared with the preset threshold values. If all the indexes meet the requirements, it indicates that the performance of the model is good and can maintain effectiveness and reliability. If there are indexes that do not meet the requirements, the reasons for the substandard indexes are analyzed, such as unreasonable model parameter settings, insufficient training samples, etc., and corresponding improvement measures are taken according to the reasons, such as retraining the model, supplementing the training samples, etc., until the performance of the model reaches the evaluation standard.

[0251] In this way, based on the generated pile load transfer simulation analysis report, suggestions can be provided for the design optimization of the pile foundation, including size adjustment, material selection, and construction process improvement of the pile foundation.

[0252] According to the analysis results of the stress and deformation characteristics of the pile foundation under different load conditions in the pile load transfer simulation analysis report, specific suggestions are provided for the design optimization of the pile foundation. In terms of pile size adjustment, if the report shows that the load is too concentrated at a certain part of the pile under a certain load, exceeding the bearing capacity of the material, it can be suggested to increase the cross-sectional size of that part to improve its bearing capacity; if the pile length is insufficient, causing excessive stress on the pile tip, it can be suggested to appropriately increase the pile length so that the load can be more evenly transferred to the deep soil.

[0253] In terms of material selection, according to the requirements for pile material performance in the report, if the existing material has a low elastic modulus, resulting in excessive deformation of the pile, it can be suggested to select a material with a higher elastic modulus; if the friction force at the pile-soil interface is insufficient, affecting the load transfer efficiency, it can be suggested to treat the surface of the pile with a material that has a higher friction coefficient or increase the surface roughness.

[0254] In terms of construction process improvement, if the report shows that the load transfer path is abnormal due to soil disturbance during construction of the pile foundation, it can be suggested to optimize the construction process, such as using more advanced pile-forming technology to reduce disturbance to the surrounding soil; if the combination of the pile and the soil is not tight enough, affecting the interface action strength, it can be suggested to improve the grouting process to improve the combination effect of the pile and the soil.

[0255] These suggestions are compiled into a design optimization report, which details the basis for the suggestions, i.e., the relevant data and conclusions in the simulation analysis report, and the benefits that may be brought about by implementing the suggestions, such as improving the bearing capacity of the pile foundation, reducing deformation, reducing engineering costs, etc., providing clear optimization directions for designers.

[0256] According to the analysis of the interaction state of the pile-soil interface in the simulation analysis report, the long-term stability of the pile foundation is evaluated, and the performance change trend of the pile foundation in the long-term use process is predicted.

[0257] According to the analysis results of the change of the interaction state of the pile-soil interface with time in the simulation analysis report, the long-term stability of the pile foundation is evaluated. The attenuation of the interaction strength of the pile-soil interface under long-term load is analyzed, as well as the influence of the above attenuation on the load transfer path and the stress state of the pile foundation.

[0258] Combined with the creep characteristics of the soil body and the fatigue performance of the pile foundation material, the performance change trend of the pile foundation in the long-term use process is predicted, such as the cumulative amount of pile displacement with time, the long-term change rule of pile-soil interface friction, etc.

[0259] If the prediction result shows that the pile foundation may have excessive deformation or bearing capacity decline in the long-term use process, which exceeds the allowable range of the project, corresponding maintenance and reinforcement suggestions are proposed, such as regular monitoring of the pile foundation, taking grouting reinforcement measures to enhance the interaction strength of the pile-soil interface, so as to ensure the long-term stable operation of the pile foundation.

[0260] Finally, the pile foundation load transfer simulation analysis is compared and analyzed with the actual engineering monitoring data, and the simulation model and analysis method are continuously improved to improve the scientificity and reliability of the pile foundation engineering design and construction.

[0261] The actual monitoring data of the pile foundation engineering in the construction and use process are collected, including the measured values of the settlement amount of the pile foundation, the stress of the pile body, the load on the pile top, etc. The collection frequency and time span of the monitoring data should match the time range of the simulation analysis.

[0262] The actual monitoring data are compared and analyzed with the prediction results in the corresponding pile foundation load transfer simulation analysis report, and the differences between them are calculated, such as the deviation of the settlement amount, the difference of the stress distribution, etc. The causes of the differences are analyzed, which may be that the values of the soil parameters in the simulation model deviate from the actual soil, the simulation of the load action conditions does not match the actual situation, the simplification of the model ignores some influencing factors, etc. According to the results of the difference analysis, the simulation model and the analysis method are improved and improved, such as adjusting the physical and mechanical parameters of the soil in the model to make them closer to the actual values, optimizing the simulation method of the load action conditions, increasing the consideration of the ignored factors in the model, etc. Through continuous comparison and feedback of the simulation analysis and the actual monitoring data, the accuracy of the simulation model and the rationality of the analysis method are gradually improved, so as to enhance the scientificity and reliability of the pile foundation engineering design and construction, and reduce the engineering risk.

[0263] Figure 2An exemplary hardware and software components of a pile load transfer simulation system 100 that can implement the idea of the present application in combination with machine learning are shown in the schematic diagram. For example, a processor 120 can be used in the pile load transfer simulation system 100 in combination with machine learning and for performing the functions in the present application.

[0264] For example, the pile load transfer simulation system 100 in combination with machine learning can include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and different forms of storage media 140, such as a disk, a ROM, or a RAM, or any combination thereof. Exemplarily, the pile load transfer simulation system 100 in combination with machine learning can also include program instructions stored in a ROM, a RAM, or other types of non-transitory storage media, or any combination thereof. The methods of the present application can be implemented according to these program instructions. The pile load transfer simulation system 100 in combination with machine learning also includes an I / O interface 150 between a computer and other input / output devices.

[0265] Further, the present application also provides a readable storage medium in which computer executable instructions are pre-set, and when a processor executes the computer executable instructions, a pile load transfer simulation method in combination with machine learning is implemented.

[0266] It should be noted that, in order to simplify the expression of the present application and to help the understanding of one or more embodiments of the present application, in the foregoing description of the embodiments of the present application, various features are sometimes incorporated into one embodiment, drawing, or description thereof.

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 simulation 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 the dynamic process curve of load transfer 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 segments 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-encoded 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 state evolution characteristic sequence of the pile-soil interface, 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 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 curves 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