A construction method for a prediction model of lateral displacement of pile foundations in high-piled wharves

Abstract

The present invention relates to the technical field of pile foundation engineering, and particularly to a method for constructing a prediction model for the lateral displacement of pile foundations in high-piled wharves. The method includes the following steps: monitoring the displacement, stress, and load parameters of the pile foundations in high-piled wharves to generate real-time monitoring parameter data; identifying the working conditions of the pile foundation structures through the real-time monitoring parameter data to generate pile foundation structure working condition data; analyzing the multi-source load characteristics of the pile foundation structure working condition data and calculating the load coupling effect to generate load action data; performing soil-pile interaction analysis based on the load action data to generate soil-pile response data; and evaluating the deterioration of material properties based on the soil-pile response data to generate material state data. The present invention establishes a working condition identification method based on multi-source monitoring data, realizes the comprehensive perception of the state of the pile foundation structure, and provides a reliable data basis for subsequent displacement prediction.

Description

Technical Field

[0001] The present invention relates to the technical field of pile foundation engineering, and particularly to a method for constructing a prediction model for lateral displacement of pile foundations in high-piled wharves. Background Art

[0002] A high-piled wharf is a common wharf structure mainly used in fields such as ports and shipping. Its design and construction processes involve multiple aspects. The piles of high-piled wharves mainly include the following types: concrete piles, steel piles, and wooden piles; the design of piles needs to consider the following factors: bearing capacity, settlement control, and seismic performance; the construction methods of piles in high-piled wharves mainly include: driving pile method, grouting method, and jacking method. The lateral displacement of pile foundations in high-piled wharves refers to the displacement that occurs along the horizontal plane under the action of external loads (such as tidal currents, waves, ship impacts, etc.). Lateral displacement is one of the important indicators for the safety and stability of pile foundations. Especially in hydraulic buildings and wharf projects, the control of lateral displacement is crucial for preventing structural damage and ensuring the long-term stability of the structure.

[0003] However, the prediction models for lateral displacement of pile foundations in high-piled wharves often have the following problems: due to the lack of analysis of the coupling effect between loads, the comprehensive influence of different loads on the lateral displacement of pile foundations cannot be fully identified, which easily causes deviations in prediction results. Traditional methods often do not consider the influence of changes in environmental factors such as temperature, humidity, and corrosion on material properties and the stability of pile foundations, affecting the deterioration assessment of materials and the reliability analysis of pile foundations. Summary of the Invention

[0004] Based on this, it is necessary for the present invention to provide a method for constructing a prediction model for lateral displacement of pile foundations in high-piled wharves to solve at least one of the above technical problems.

[0005] To achieve the above object, a method for constructing a prediction model for lateral displacement of pile foundations in high-piled wharves includes the following steps:

[0006] Step S1: Monitor the displacement, stress, and load parameters of the pile foundations in high-piled wharves to generate real-time monitoring parameter data; identify the working conditions of the pile foundation structure through the real-time monitoring parameter data to generate pile foundation structure working condition data;

[0007] Step S2: Analyze the multi-source load characteristics of the pile foundation structure working condition data and calculate the load coupling effect to generate load action data; analyze the soil-pile interaction based on the load action data to generate soil-pile response data;

[0008] Step S3: Conduct a material property degradation assessment based on the soil-pile response data to generate material status data; conduct an analysis of the influence of environmental factors such as temperature, humidity, and corrosion based on the material status data to generate environmental impact characteristic data; extract features from the real-time monitoring parameter data, and perform dynamic feature fusion based on correlation analysis on the environment-corrected data to obtain the displacement-environment-load correlation characteristic data;

[0009] Step S4: Use the correlation characteristic data to train a deep learning model to obtain prediction model parameters; conduct a time series feature analysis on the load action data and the prediction model parameters, and perform prediction accuracy optimization based on the attention mechanism to generate dynamic prediction control data; adaptively update the prediction model parameters according to the dynamic prediction control data to obtain optimized model parameters;

[0010] Step S5: Continuously predict the lateral displacement of the pile foundation using the optimized model parameters to generate displacement prediction data; conduct uncertainty quantification and reliability assessment on the displacement prediction data to generate corrected prediction result data.

[0011] The present invention monitors displacement, stress, and load parameters in real time, which helps to quickly identify the structural state of pile foundations and generate accurate monitoring data. These data can reflect the behavior of the structure under actual working conditions and provide a data basis for subsequent predictions. By identifying the working conditions through the monitoring data, the structural state can be analyzed and classified in real time to obtain the characteristics of the working conditions. This characteristic provides reliable reference data for subsequent load coupling and soil-pile analysis and improves the adaptability of the model. The analysis of multi-source load characteristics combined with the calculation of coupling effects can effectively capture the multiple effects of external loads on pile foundations and provide more accurate load data for the analysis of soil-pile interaction. The soil-pile interaction analysis can be refined by combining the actual working conditions, and the generated soil-pile response data can accurately describe the response behavior of the structure under complex working conditions. This is crucial for the model to capture the synergistic relationship between the soil and the pile and improve the prediction accuracy. The assessment of material property degradation captures the degradation process of structural materials under different loads by analyzing the soil-pile response data. This helps to identify potential fatigue damage and supports long-term performance prediction. The analysis of the influence of environmental factors generates environmental impact characteristic data by evaluating temperature, humidity, and corrosion, further improving the adaptability of the model in the actual environment. Dynamically reflecting the influence of environmental factors on the structure in the characteristic data improves the accuracy and wide applicability of the model prediction. By training a deep learning model by correlating the characteristic data, the model can autonomously identify the complex relationships in the data and improve the prediction accuracy. Based on the optimization of time series feature analysis and attention mechanism, the sensitivity of the model to future time series changes can be enhanced, improving the timeliness and stability of the prediction. The dynamic prediction control data adaptively updates and optimizes the model parameters, enabling the model to quickly adjust when the load and environmental factors change, ensuring the long-term stability and accuracy of the model and reducing the cumulative effect of prediction errors. After optimizing the model parameters, continuous prediction of the lateral displacement of the pile foundation is realized, achieving a dynamic assessment of the displacement behavior. By quantifying uncertainty and evaluating reliability, uncertain factors in the prediction can be identified and the prediction results can be corrected in a timely manner. The corrected prediction result data makes the error control in the actual application of the model more accurate, improves the prediction reliability and practicality of the model, and enhances the stability and effectiveness of the model under complex working conditions and environments. Generally speaking, through systematic steps and data fusion, this method realizes the accurate prediction of the lateral displacement of pile foundations of high-pile wharves, demonstrating significant technical advantages in working condition identification, soil-pile interaction analysis, material degradation, environmental impact, and prediction model optimization, ensuring the practicality and reliability of the method. Brief Description of the Drawings

[0012] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments read with reference to the accompanying drawings:

[0013] Figure 1Schematic diagram of the step process for constructing the prediction model of the lateral displacement of the pile foundation of a high-piled wharf according to the present invention;

[0014] Figure 2 is Figure 1 detailed schematic diagram of step S1 in

[0015] Figure 3 is Figure 1 detailed schematic diagram of step S2 in Specific implementation manner

[0016] The technical method of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0017] In addition, the accompanying drawings are only schematic diagrams of the present invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and thus their repeated description will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0018] It should be understood that although the terms "first", "second", etc. may be used here to describe each unit, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used here includes any and all combinations of one or more of the listed related items.

[0019] To achieve the above object, please refer to Figures 1 to 3 , the present invention provides a method for constructing a prediction model of the lateral displacement of the pile foundation of a high-piled wharf, and the method includes the following steps:

[0020] Step S1: Monitor the displacement, stress and load parameters of the pile foundation of the high-piled wharf to generate real-time monitoring parameter data; identify the working conditions of the pile foundation structure through the real-time monitoring parameter data to generate pile foundation structure working condition data;

[0021] In the embodiments of the present invention, various sensors are arranged on the pile foundation to collect displacement, stress, and load parameter data in real time. A data acquisition system is used to continuously record these data to generate real-time monitoring parameter data, reflecting the dynamic working conditions of the structure. A structural working condition recognition algorithm (such as a Bayesian classifier) is used to analyze these parameter data to identify the working condition state of the pile foundation structure. Through data processing, the structural conditions of the pile foundation under different loads and environmental conditions are classified, thereby generating pile foundation structure working condition data, which provides early warning for scenarios with frequent tidal changes or variable sediment compositions.

[0022] Step S2: Conduct multi-source load characteristic analysis on the pile foundation structure working condition data, and calculate the load coupling effect to generate load action data; conduct soil-pile interaction analysis based on the load action data to generate soil-pile response data;

[0023] In the embodiments of the present invention, multi-source load characteristic analysis is conducted on the pile foundation structure working condition data, including load types such as tidal force, ship impact force, and thermal stress. The finite element analysis (FEM) is used to calculate the coupling effect of the loads to obtain the load action data, revealing the influence of the combined action of multiple loads on the pile foundation. Based on the load action data, the improved Mohr-Coulomb model is applied for soil-pile interaction analysis to calculate the soil response under each load scenario, thereby generating soil-pile response data. This process is particularly important for high-pile wharves with large water level changes because of the complex load interactions.

[0024] Step S3: Conduct material performance degradation assessment based on the soil-pile response data to generate material state data; conduct environmental factor influence analysis based on temperature, humidity, and corrosion on the material state data to generate environmental influence characteristic data; extract features from the real-time monitoring parameter data, and conduct dynamic feature fusion of the environmental correction data based on correlation analysis to obtain the displacement-environment-load correlation feature data;

[0025] In the embodiments of the present invention, the soil-pile response data is used to evaluate the material performance degradation. The aging model is used to simulate the performance of the material changing with time under load and environmental stress to generate material state data. Combining the influences of temperature, humidity, and corrosion, the environmental influence factor is calculated and dynamically adjusted through real-time environmental monitoring input. The displacement features of the real-time monitoring data are extracted, and a feature fusion method based on correlation analysis is used to fuse the environmental correction data with the displacement feature data to generate the displacement-environment-load correlation feature data. This process is particularly crucial in tropical ports with high humidity because the rapid degradation of materials will directly affect the stability of the pile foundation.

[0026] Step S4: Use the associated feature data to train a deep learning model to obtain prediction model parameters; perform time-series feature analysis on the load action data and the prediction model parameters, and optimize the prediction accuracy based on the attention mechanism to generate dynamic prediction control data; adaptively update the prediction model parameters according to the dynamic prediction control data to obtain optimized model parameters;

[0027] In an embodiment of the present invention, the associated feature data is used to train a deep learning model, such as a long short-term memory (LSTM) network, to capture the time dependence in the data and generate model parameters for predicting the lateral displacement of the pile foundation. The time-series feature analysis is performed on the load action data in combination with these model parameters, and the attention mechanism is applied to improve the prediction accuracy to generate dynamic prediction control data. The adaptive parameter update algorithm continuously optimizes the model parameters based on the control data, thereby forming optimized model parameters applicable to scenarios with frequent load changes such as seasonal storms or changes in port traffic flow.

[0028] Step S5: Continuously predict the lateral displacement of the pile foundation using the optimized model parameters to generate displacement prediction data; perform uncertainty quantification and reliability assessment on the displacement prediction data to generate corrected prediction result data.

[0029] In an embodiment of the present invention, the optimized model parameters are used to continuously predict the lateral displacement of the pile foundation to generate displacement prediction data. Monte Carlo simulation is used for uncertainty quantification to evaluate the reliability of the prediction and calculate the confidence interval for each prediction. The reliability assessment is performed on the prediction data to generate corrected prediction result data. This implementation process allows the port management department to evaluate the displacement prediction of the pile foundation under complex load and environmental conditions, ensuring that the prediction meets the reliability threshold, and is particularly applicable to commercial ports with high load demands and significant environmental changes.

[0030] The present invention monitors displacement, stress, and load parameters in real time, which helps to quickly identify the structural state of pile foundations and generate accurate monitoring data. These data can reflect the behavior of the structure under actual working conditions and provide a data basis for subsequent predictions. By identifying working conditions through the monitoring data, the structural state can be analyzed and classified in real time to obtain the characteristics of the working conditions. This characteristic provides reliable reference data for subsequent load coupling and soil-pile analysis and improves the adaptability of the model. The multi-source load characteristic analysis combined with the coupling effect calculation can effectively capture the multiple effects of external loads on pile foundations and provide more accurate load data for the analysis of soil-pile interaction. The soil-pile interaction analysis can be refined by combining the actual working conditions, and the generated soil-pile response data can accurately describe the response behavior of the structure under complex working conditions. This is crucial for the model to capture the synergistic relationship between the soil and the pile and improve the prediction accuracy. The evaluation of material property degradation captures the degradation process of structural materials under different loads by analyzing the soil-pile response data. This helps to identify potential fatigue damage and supports long-term performance prediction. The analysis of the influence of environmental factors generates environmental impact characteristic data by evaluating temperature, humidity, and corrosion, further improving the adaptability of the model in the actual environment. Dynamically reflecting the influence of environmental factors on the structure in the characteristic data improves the accuracy and wide applicability of model prediction. By training a deep learning model by correlating the characteristic data, the model can autonomously identify the complex relationships in the data and improve the prediction accuracy. Based on the optimization of time series feature analysis and attention mechanism, the sensitivity of the model to future time series changes can be enhanced, improving the timeliness and stability of prediction. The dynamic prediction control data adaptively updates and optimizes the model parameters, enabling the model to quickly adjust when the load and environmental factors change, ensuring the long-term stability and accuracy of the model and reducing the cumulative effect of prediction errors. After optimizing the model parameters, continuous prediction of the lateral displacement of the pile foundation is carried out to achieve dynamic evaluation of the displacement behavior. Through uncertainty quantification and reliability assessment, the uncertain factors in the prediction can be identified and the prediction results can be corrected in a timely manner. The corrected prediction result data enables more accurate error control in the actual application of the model, improves the prediction reliability and practicality of the model, and enhances the stability and effectiveness of the model under complex working conditions and environments. Generally speaking, through systematic steps and data fusion, this method achieves accurate prediction of the lateral displacement of pile foundations of high-pile wharves, demonstrating significant technical advantages in working condition identification, soil-pile interaction analysis, material degradation, environmental impact, and prediction model optimization, ensuring the practicality and reliability of the method.

[0031] Preferably, step S1 includes the following steps:

[0032] Step S11: Use multi-source sensors installed on the pile foundation of the high-piled wharf and the surrounding soil to collect sensor parameters in real time, so as to obtain pile foundation displacement data, pile foundation force data, and environmental load data, where the multi-source sensors include displacement sensors, stress sensors, and load sensors;

[0033] In the embodiment of the present invention, in order to obtain real-time pile foundation displacement, stress, and environmental load data, multi-source sensors are installed in the pile foundation of the high-piled wharf and the surrounding soil area, including high-precision displacement sensors, stress sensors, and load sensors. The displacement sensors are installed at multiple positions on the top, bottom, and middle of the pile foundation to capture displacement conditions at different heights; the stress sensors are installed on the surface of the pile foundation and in the soil to monitor the stress distribution at different depths of the pile foundation; the load sensors are installed in the soil around the pile foundation to evaluate the load changes under the action of the surrounding soil and water. In practical applications, for example, in coastal areas, due to the influence of hydrological conditions, the sampling frequency of the sensors needs to be set to 10Hz to capture instantaneous load and displacement changes and ensure that the data collected by the sensors in real time reflects the dynamic changes of the wharf environment.

[0034] Step S12: Perform data preprocessing on the pile foundation displacement data, pile foundation force data, and environmental load data, and perform data dimension unification processing to obtain real-time monitoring parameter data;

[0035] The pile foundation displacement, force, and environmental load data collected in the embodiment of the present invention may have outliers or missing values. In order to ensure the accuracy and consistency of the data, first perform abnormal data cleaning and complementation processing, and then use a standardization method to unify data with different dimensions such as displacement (unit: millimeter), stress (unit: MPa), and load (unit: kN) into dimensionless values. Through normalization or Z-score standardization, all data are within the same range to facilitate subsequent data analysis. For example, for the abnormal high-frequency load data generated by a certain port in a specific season, the 3σ criterion can be used to remove outliers, so as to obtain stable real-time monitoring parameter data for further analysis.

[0036] Step S13: Perform static condition identification on the real-time monitoring parameter data to obtain static condition characteristic data, where the static condition identification specifically includes static load condition classification, structural deformation characteristic extraction, stress distribution pattern identification, and support condition evaluation;

[0037] The static condition identification in the embodiments of the present invention is mainly based on the low-frequency part of the monitoring data. First, the data is subjected to low-pass filtering to remove the high-frequency dynamic response, ensuring that only static features are obtained. Then, the real-time monitoring data is classified by working conditions, and the static load response characteristics of the pile foundation under specific loading conditions are extracted. Taking a seawall wharf as an example, the static load characteristics can be identified by comparing the stress distribution of the pile foundation when there is no tidal influence and the wind is stable, and the structural deformation analysis is carried out. At the same time, by analyzing the stress distribution pattern, it is possible to identify whether there is abnormal stress in local areas of the wharf pile foundation, and classify the support conditions to ensure the stability of the wharf under normal conditions.

[0038] Step S14: Perform dynamic condition identification of extracting dynamic response characteristics, vibration mode analysis, dynamic load characteristic identification and structural dynamic characteristic evaluation according to the real-time monitoring parameter data, so as to obtain dynamic condition characteristic data;

[0039] For the dynamic condition identification in the embodiments of the present invention, high-frequency dynamic response characteristics are mainly extracted. The vibration frequency distribution is obtained by fast Fourier transform (FFT), and the vibration mode of the wharf pile foundation is identified. Then, based on high-pass filtering, the low-frequency static load signal is removed, and the high-frequency response is retained for analyzing the instantaneous load characteristics. In actual operation, for example, in the case of a storm or a ship berthing, the dynamic response characteristics of the pile foundation structure are identified by modal analysis to ensure its stability under sudden loads. In addition, by analyzing the dynamic load characteristic data, the seismic performance of the wharf under different weather conditions can be judged, providing a basis for subsequent disaster prevention design.

[0040] Step S15: Perform information collaborative fusion processing on the static condition characteristic data and the dynamic condition characteristic data, so as to obtain the pile foundation structure condition data.

[0041] In the embodiments of the present invention, the static condition characteristic data and the dynamic condition characteristic data are subjected to information fusion processing. First, the two types of data are subjected to dimensionality reduction processing by principal component analysis (PCA) to remove redundant information and extract the main features. Then, Kalman filtering is used for collaborative processing to obtain more accurate pile foundation structure condition data. In practical applications, for example, when evaluating the overall wind and wave resistance performance of a coastal wharf, fusing the static deformation characteristics and the dynamic vibration characteristics to obtain more representative pile foundation overall structure condition data, which can more comprehensively reflect the stability of the wharf under various working conditions and provide a basis for subsequent structural safety assessment.

[0042] The present invention utilizes multi-source sensors (including displacement, stress, and load sensors) to collect key parameters of pile foundations and surrounding soil in real time, ensuring the comprehensiveness of data. The multi-source sensors can synchronously capture displacement, stress, and environmental load information, making the acquired data more real and complete, and fully reflecting the state of the pile foundation under actual working conditions. This real-time data acquisition method enhances the timeliness of data, contributes to the dynamic response of subsequent working condition identification and prediction, and makes the model more sensitive to changes in working conditions. Preprocessing the data, including steps such as denoising and filling in missing data, ensures the quality of the data and improves the reliability of the data. Unified processing of data dimensions ensures the comparability and consistency of numerical values from different data sources, making subsequent working condition identification and feature analysis more accurate. The preprocessed data reflects the real monitoring situation with higher precision, provides high-quality basic data for subsequent analysis, and reduces the deviation risk of the model caused by data errors. Static working condition identification extracts characteristic data of the pile foundation under static load, such as static load condition classification, structural deformation characteristics, stress distribution patterns, and support condition evaluation, etc. These static characteristics can reflect the stress state of the pile foundation under stable conditions, especially helping to understand the structural performance under long-term bearing conditions. The extraction of static characteristic data plays an important role in setting parameters of the prediction model and analyzing the long-term stress behavior of the pile foundation, providing basic state parameters for subsequent working condition analysis. Dynamic working condition identification captures the vibration modes, dynamic load characteristics, etc. of the pile foundation under dynamic load through the extraction of dynamic characteristics of real-time monitoring parameter data. These dynamic characteristic data can reflect the response of the structure under vibration and periodic loads, and are important data for the dynamic safety assessment and model training of the pile foundation. By evaluating the dynamic characteristics of the structure, the safety state of the pile foundation under different dynamic working conditions can be predicted, providing a reference for risk prediction under dynamic conditions, and enabling the model to adapt to complex dynamic working condition changes. Performing information fusion processing on the static and dynamic working condition characteristic data helps to form comprehensive pile foundation working condition data. The fused working condition data contains the comprehensive characteristics of the structure under static and dynamic loads, fully reflecting the overall stress and deformation of the pile foundation. Collaborative fusion processing enhances the correlation between data, helps to identify the process of working condition conversion, and provides a more accurate working condition reference for subsequent lateral displacement prediction. This data fusion improves the overall adaptability of the prediction model and ensures the accuracy and reliability of the prediction. Generally speaking, this process establishes an efficient multi-source sensor data acquisition, data preprocessing, and working condition identification method, realizing full-process support from data acquisition to working condition feature extraction. Through the collaborative fusion of static and dynamic working condition characteristic data, the model can comprehensively understand the structural characteristics and stress state of the pile foundation from multiple dimensions, improving the robustness and adaptability of the prediction model under complex working conditions.

[0043] Preferably, step S13 includes the following steps:

[0044] Step S131: Classify the environmental load data in the real-time monitoring parameter data according to the loading mode, loading amplitude, and loading sequence, and identify the load conditions according to the preset static load condition classification criteria, so as to obtain the static load condition classification data;

[0045] In the classification process of the environmental load data in the embodiments of the present invention, the loading mode, loading amplitude, and loading sequence are first analyzed. For example, for the load data collected during a typhoon, by analyzing the hourly loading amplitude change curve, the intensity increase of the wind force within a short period of time can be identified, and a typical wind load pattern can be classified. At the same time, according to the preset static load condition classification criteria, the loading characteristics of a smaller amplitude and a stable sequence are classified as static conditions, so as to identify different static load types, such as the constant wind pressure acting for a long time or the water pressure under the influence of tides, and generate the static load condition classification data. Such data can effectively reflect the bearing state of the wharf pile foundation over a long period of time and help to judge the long-term stability of the structure.

[0046] Step S132: Extract the deformation characteristics of the pile foundation displacement data, and calculate the characteristic quantities of the displacement amplitude, deformation trend, deformation rate, and deformation mode, so as to obtain the structural deformation characteristic data;

[0047] In the embodiments of the present invention, to extract the deformation characteristics of the pile foundation, the maximum and minimum amplitudes are first obtained from the displacement data to calculate the deformation amplitude. At the same time, based on time series analysis, the deformation rate is obtained and the trend is identified. In a typical application, for example, when a wharf pile foundation is deformed under the influence of water pressure, the periodic change of the deformation with the tides can be observed through the daily displacement amplitude records. Then, the acceleration curve is used to analyze the deformation rate to obtain the velocity characteristics of the deformation. At the same time, the deformation mode is identified by the method of curve fitting to judge whether the pile foundation is bent upward, horizontally displaced, or settled downward, so as to obtain the structural deformation characteristic data. These data can be directly used for subsequent structural safety assessment to help predict the long-term deformation trend of the pile foundation.

[0048] Step S133: Identify the stress distribution pattern based on the pile foundation force data according to the stress distribution law, stress concentration area, stress transfer path, and stress level, so as to obtain the stress distribution pattern data;

[0049] When identifying the stress distribution of the pile foundation in the embodiments of the present invention, first, the stress concentration area and stress level are analyzed based on the three-dimensional stress distribution diagram of the pile foundation force data. For example, in the data collected under high wind speeds, it can be observed that obvious stress concentration phenomena occur near the upper end of the pile foundation, and at the same time, the stress transmission path in the pile body shows a distribution law of gradually decaying from the pile top to the bottom end of the pile body. Then, according to the change of the stress level, the stress distribution characteristics of the key parts are identified and it is determined whether there is a high stress concentration phenomenon. The stress distribution pattern data generated in this way can be used to detect potential structural weak areas and judge the stress-bearing capacity of the pile foundation under harsh working conditions.

[0050] Step S134: Evaluate the interaction relationship between the pile foundation and the soil body according to the pile foundation displacement data and the pile foundation force data, and generate constraint conditions for the pile end constraint, soil body constraint, and pile body constraint of the pile foundation, so as to obtain the support condition data;

[0051] When analyzing the interaction between the pile foundation and the soil body based on the pile foundation displacement and force data in the embodiments of the present invention, first, the vertical deformation of the bottom end of the pile foundation is evaluated to generate the constraint condition data of the pile end. By monitoring the deformation behavior of the pile foundation under high loads, the support strength of the soil body is determined and used as the data source of the soil body constraint. At the same time, according to the distribution of the stress transmission in the pile body, the horizontal constraint condition of the pile body is judged to identify the strength of the pile foundation being constrained by the surrounding soil body in the horizontal direction. For example, in soft soil areas, the soil body constraint is relatively weak, while in sandy soil areas, the constraint is relatively strong. The support condition data can provide guidance for the subsequent pile foundation structure design to ensure a reasonable analysis of the interaction between the pile foundation and the soil body.

[0052] Step S135: Perform feature fusion on the static load condition classification data, structural deformation characteristic data, stress distribution pattern data, and support condition data to obtain the static condition characteristic data.

[0053] When the embodiments of the present invention perform feature fusion on the static load condition classification data, structural deformation characteristic data, stress distribution pattern data, and support condition data, first, the different feature data are standardized by the weighted average method, then the feature vectors of various data are extracted, and dimensionality reduction fusion is performed through principal component analysis (PCA). Finally, the feature data representing the overall static condition is obtained. For example, after integrating various features, it can be clearly seen that under static conditions, the pile foundation structure is mainly affected by long-term tidal loads, and its deformation trend is stable and there is no stress concentration phenomenon. The static condition characteristic data provides an important basis for evaluating the structural stability of the pile foundation under normal loads and supports the decision-making of structural maintenance and safety inspections.

[0054] The present invention classifies environmental load data according to the loading mode, amplitude, and time series, and identifies working conditions through the static load condition standard, obtaining static load condition classification data. This step distinguishes the action characteristics of different types of environmental loads, enabling the model to identify the load effects under specific static conditions. The generation of static load classification data provides a reference basis for subsequent analysis and can effectively improve the prediction accuracy of pile foundation responses under different environmental conditions. By analyzing the pile foundation displacement data to extract deformation characteristics, including displacement amplitude, deformation trend, deformation rate, and mode, etc., it helps to understand in detail the deformation behavior of the pile foundation under static loads. These characteristic data can accurately reflect the actual deformation of the pile foundation under static working conditions and provide a comprehensive description of the structural deformation characteristics. The deformation characteristic data can help identify the response law of the pile foundation under specific load conditions, facilitate the subsequent analysis of the impact of deformation on the safety and stability of the pile foundation, and provide reliable data support for displacement prediction. By identifying the stress distribution pattern based on the pile foundation force data, characteristics such as stress concentration areas, transmission paths, and stress levels are extracted. The stress distribution pattern can reflect the stress state of the pile foundation structure under static working conditions and provide a reliable basis for judging the key stress areas and possible stress concentration points of the pile foundation. The stress distribution pattern data helps to identify the stress characteristics of the pile foundation, optimize the pile foundation structure design and construction process, and effectively reduce the structural risks caused by stress concentration. According to the pile foundation displacement and force data, the soil-pile interaction analysis is carried out to generate the constraint conditions of the pile tip, soil body, and pile body. This process identifies the mutual influence between the pile foundation and the surrounding soil, comprehensively evaluates the force conditions and support conditions of the entire pile foundation, and ensures that the model can truly reflect the support characteristics of the structure. The support condition data provides accurate boundary conditions for the displacement prediction model, helps the model to accurately simulate under specific loads, and makes the prediction closer to the actual working conditions. By fusing the static load condition classification, deformation characteristics, stress distribution pattern, and support condition data, comprehensive static working condition characteristic data is formed. Through feature fusion, the model can obtain more comprehensive and consistent working condition information, improving the correlation between data. The static working condition characteristic data lays a solid foundation for the training of subsequent dynamic prediction models, enabling the model to achieve higher accuracy under static conditions and optimizing the stability and adaptability of the prediction results. Generally speaking, these steps form multi-dimensional characteristic data covering load, deformation, stress, and support conditions through detailed static characteristic analysis, providing detailed static data support for the model. This data processing flow improves the prediction accuracy of the model under static working conditions, and through feature fusion, the model has stronger adaptability to complex working conditions, which helps to further optimize the reliability of pile foundation lateral displacement prediction.

[0055] Preferably, step S14 includes the following steps:

[0056] Step S141: Extract the dynamic response characteristics of the pile foundation structure for the dynamic response parameters based on displacement time history, acceleration response, and stress fluctuation from the real-time monitoring parameter data, so as to obtain the dynamic response characteristic data;

[0057] In the embodiment of the present invention, in order to extract the dynamic response characteristics of the pile foundation structure, time history analysis is first performed on the real-time monitored displacement, acceleration, and stress data. Using the acceleration time history data in the time domain can reveal the dynamic response of the pile foundation under specific working conditions, such as the instantaneous acceleration peak value under the impact of ocean waves and the vibration recovery time of the pile foundation. By performing Fourier transform analysis on the spectral characteristics of stress fluctuations, the main frequency components induced by dynamic loads are identified, so as to determine the dynamic response characteristic data. For example, when the pile foundation is located in a coastal area and is subjected to periodic ocean wave impacts, significant response characteristics in the frequency range of 1 Hz to 5 Hz can be detected, and this data is used to analyze the structural stability of the pile foundation under different dynamic load conditions.

[0058] Step S142: Perform vibration mode analysis based on the dynamic response characteristic data, so as to obtain the vibration mode characteristic data;

[0059] The vibration mode analysis in the embodiment of the present invention utilizes the dynamic response characteristic data to perform vibration mode analysis on the pile foundation structure through a modal analysis software such as Ansys to obtain the vibration mode characteristic data. First, the structural model of the pile foundation is imported into the software, the boundary conditions are set, and the load distribution of the typical working conditions is applied to the pile foundation. Through eigenvalue analysis, its natural frequency and vibration mode are extracted. For example, for a 30-meter-high pile, in the vibration mode analysis, it can be observed that its first-order modal frequency is 1.8 Hz and the second-order modal frequency is about 4.3 Hz, corresponding to different vibration forms such as lateral swing and vertical deformation. Through these characteristic data, engineers can judge the seismic and wind resistance capabilities of the pile foundation to optimize the structural design and improve the stability.

[0060] Step S143: Extract the dynamic load characteristics of wave load, wind load, and seismic load from the environmental load data, and perform parameter statistics on the amplitude, frequency, phase, and action direction of the dynamic load characteristics, so as to obtain the dynamic load characteristic data;

[0061] In an embodiment of the present invention for extracting dynamic load characteristic data, first, the wave, wind load, and seismic load included in the environmental load data are classified and analyzed. Through the amplitude and frequency distribution of the wave load, the impact intensity and frequency of the sea wave are analyzed; for the wind load, the acting amplitude and direction on the structure are extracted through the statistics of the wind speed and wind direction data; and for the seismic load, the phase difference, amplitude, and frequency of the seismic action are extracted through the seismic acceleration and stress waveform records. For example, in a certain coastal high-piled wharf, the average amplitude of the sea wave load is 5 kN, and the frequency is about 0.7 Hz, while the wind load direction is stable at an angle of 30°, and the peak value is 20 kN. This characteristic data provides key input parameters for dynamic response analysis and structural design.

[0062] Step S144: Evaluate the dynamic characteristic parameters based on the dynamic response characteristic data and the vibration mode characteristic data to obtain the structural dynamic characteristic data, where the dynamic characteristic parameters include the stiffness characteristic, mass distribution, and energy dissipation of the pile foundation structure;

[0063] In an embodiment of the present invention, based on the dynamic response characteristics and vibration mode characteristic data, the dynamic characteristic parameters of the pile foundation structure are evaluated, including the stiffness characteristic, mass distribution, and energy dissipation. By analyzing the vibration frequency and modal vibration mode of the pile foundation, the structural stiffness can be evaluated. Using the detection of energy dissipation in the vibration experiment, the durability of the pile foundation under dynamic loads can be judged. Assuming that the first-mode vibration frequency of the pile foundation under seismic load is 2 Hz, and its energy dissipation rate is detected to be 5% through amplitude decay, it shows that the pile foundation has a high seismic resistance during vibration, thereby obtaining the structural dynamic characteristic data. These data can further optimize the pile foundation structure and improve its durability and seismic performance.

[0064] Step S145: Conduct a characteristic correlation analysis on the dynamic response characteristic data, vibration mode characteristic data, dynamic load characteristic data, and structural dynamic characteristic data to obtain the dynamic condition characteristic data.

[0065] In an embodiment of the present invention for generating the dynamic condition characteristic data, the dynamic response, vibration mode, dynamic load, and structural dynamic characteristic data are correlated and analyzed. Through the correlation analysis, the response characteristics of the pile foundation under different load conditions can be revealed. Using a multiple regression model, the modal characteristics, response amplitudes, frequencies under different conditions are correlated with the corresponding dynamic load characteristic data to identify the pile foundation response under specific load combinations. For example, when the wind load and seismic load act in combination, by comparing the displacement responses of the pile foundation in each mode, it can be found that the amplitude increase rate during the combined action reaches 30%, which is helpful for the extraction of dynamic condition characteristics and provides decision support for the real-time monitoring and dynamic control of the pile foundation structure.

[0066] The present invention extracts the displacement time history, acceleration response, and stress fluctuation from the real-time monitoring data to obtain the dynamic response characteristics of the pile foundation structure. This step provides the true response of the pile foundation structure under dynamic conditions, revealing the dynamic behavior of the pile foundation under different dynamic loads. The dynamic response characteristic data provides detailed basic data for subsequent vibration mode analysis and dynamic characteristic evaluation, ensuring the model's ability to capture dynamic responses and helping to improve the dynamic prediction accuracy of the model. Through the vibration mode analysis of the dynamic response characteristic data, vibration mode characteristic data is obtained, which helps to identify the natural frequency, vibration mode, damping ratio, and other characteristics of the pile foundation structure. These characteristics are important indicators of dynamic response, reflecting the interaction relationship between the vibration characteristics of the structure and external loads. The vibration mode characteristic data enables the model to better understand the dynamic response of the pile foundation structure at different frequencies, ensuring the prediction accuracy of the model under various dynamic conditions, especially providing more reliable predictions under complex fluctuating load conditions. By extracting the characteristics of dynamic loads such as waves, wind loads, and earthquakes from the environmental load data, detailed parameters of these loads in terms of amplitude, frequency, phase, and acting direction are obtained. This process provides an accurate description of various complex loads for the model, enhancing the model's adaptability under multi-source loads. The generation of dynamic load characteristic data can ensure that the model better simulates the action of various dynamic loads under real conditions, effectively improving the response accuracy of the model under fluctuating and sudden loads. Combining the dynamic response characteristic data and the vibration mode characteristic data, the dynamic characteristic parameters such as the stiffness characteristics, mass distribution, and energy dissipation of the pile foundation structure are evaluated. The dynamic characteristic data describes the stability and resistance ability of the pile foundation structure under dynamic action, providing the necessary stiffness and energy dissipation indicators for the prediction model. The structural dynamic characteristic data not only supports dynamic displacement prediction but also can evaluate the durability and safety of the structure under different load conditions, helping to more accurately predict the long-term performance of the structure under dynamic load conditions. The dynamic response characteristic data, vibration mode characteristic data, dynamic load characteristic data, and structural dynamic characteristic data are subjected to correlation analysis to form dynamic condition characteristic data. Through feature correlation, the model can synthesize multiple dynamic features, thereby more accurately capturing the overall response of the pile foundation under dynamic conditions. The formation of dynamic condition characteristic data ensures the model's comprehensive response ability to diverse dynamic conditions, enabling it to provide more accurate and reliable displacement prediction results under complex conditions. Generally speaking, these steps effectively improve the model's adaptability and prediction accuracy to complex dynamic environments through the comprehensive extraction and analysis of the dynamic condition characteristics of the pile foundation structure. This series of feature extraction and analysis lays a solid foundation for the dynamic response part of the pile foundation lateral displacement prediction model, ensuring the accuracy and strong reliability of the displacement prediction results under different conditions.

[0067] Preferably, step S142 includes the following steps:

[0068] Step S1421: Extract the natural frequencies of the pile foundation structure from the dynamic response characteristic data to obtain natural frequency characteristic data;

[0069] In the embodiment of the present invention, in order to extract the natural frequencies of the pile foundation structure, first perform spectral analysis on the dynamic response characteristic data, and use the fast Fourier transform (FFT) technology to convert the time-domain data into frequency-domain data, so as to obtain the response amplitudes of the pile foundation at different frequencies. Identify the natural frequencies by finding the positions of significant frequency peaks in the spectrogram. For example, in the pile foundation structure of a coastal high-piled wharf, obvious peaks appear at frequencies of 1.5 Hz and 3.8 Hz through FFT analysis, representing the first and second natural frequencies of the pile foundation. These data provide important natural frequency characteristic data for subsequent modal analysis.

[0070] Step S1422: Perform modal shape analysis based on the natural frequency characteristic data, extract the vibration modal shapes of the pile foundation structure, and calculate the amplitudes and phase characteristics of each order of mode, so as to obtain modal shape data;

[0071] In the embodiment of the present invention, based on the extracted natural frequency characteristic data, perform modal shape analysis on the pile foundation structure. First, input the structural model into modal analysis software, such as Ansys or Matlab, apply a small simulated load at the natural frequency, and extract the modal shape by calculating the displacement response of the structure. Calculate the amplitude and phase characteristics of each order of mode in the analysis. For example, in the first-order mode at 2 Hz, the amplitude of the top of the pile foundation is 0.005 m, and there is a 180° phase difference relative to the base. These modal shape data can intuitively show the vibration behavior of the pile foundation in different modes, providing data support for structural health assessment and improved design.

[0072] Step S1423: Calculate the damping ratios of each order of mode of the pile foundation structure based on the semi-power method for the dynamic response characteristic data and the modal shape data, so as to obtain damping characteristic data;

[0073] In the embodiment of the present invention, in order to calculate the damping ratios of each order of mode of the pile foundation structure, the semi-power method is adopted, and the damping is calculated by analyzing the frequency bandwidth characteristics of the dynamic response characteristic data and the modal shape data. First, select the natural frequency of a specific mode, and then record the peak frequency and half-power point frequencies of the response curve near this frequency. Calculate the damping ratio using the formula. For example, for the first-order mode, the peak value of the response amplitude is 0.003 m at 1.5 Hz, and the half-power point frequencies are 1.4 Hz and 1.6 Hz respectively. The calculated damping ratio is 3%. This damping characteristic data can reflect the vibration energy dissipation ability of the structure and provide an important reference for seismic design optimization.

[0074] Step S1424: Analyze the changing rules of the natural frequency characteristic data, the modal vibration shape data and the damping characteristic data based on the time series, and evaluate the time-varying characteristics of the dynamic characteristics of the pile foundation structure, so as to obtain the vibration modal characteristic data.

[0075] The embodiment of the present invention performs time-varying characteristic analysis based on natural frequency, modal vibration shape and damping characteristic data, and evaluates the changing trend of the dynamic characteristics of the pile foundation structure over time through the time series analysis method. The vibration modal parameters are modeled using an autoregressive model (AR) to observe the changes in natural frequency and damping ratio at different time points. For example, long-term monitoring results show that after 6 months, the first-order natural frequency dropped from 1.5Hz to 1.48Hz, and the damping ratio increased from 3% to 3.2%, reflecting possible minor damage or performance degradation to the structure. Through this time-varying characteristic evaluation, potential damage to the structure can be warned in advance, providing data support for maintenance decisions.

[0076] The present invention extracts the natural frequencies from the dynamic response characteristic data to obtain natural frequency characteristic data, which are the basic parameters of structural vibration. The natural frequency reflects the free vibration characteristics of the structure without external excitation and can reveal the overall stiffness and mass distribution of the structure. Extracting the natural frequency data enables the model to accurately identify the natural vibration characteristics of the pile foundation structure, helps predict the response of the pile foundation under specific frequency excitations, and provides a basis for subsequent modal analysis. Based on the natural frequency characteristic data, the vibration modal shapes of the pile foundation structure are analyzed, including the vibration amplitude and phase characteristics, to obtain the modal shape data of each order. The modal shape data reveals the deformation modes and relative displacement distributions of the structure under different modes, which is particularly crucial for identifying the deformation characteristics of the structure under different working conditions. Modal shape analysis not only enhances the model's understanding of the dynamic behavior of the pile foundation structure but also provides vibration amplitude and phase data, which can be used to predict the deformation trend and dynamic stability of the pile foundation under external loads. The half-power method is used to calculate the damping ratio from the dynamic response characteristic data and the modal shape data to obtain the damping characteristic data. The damping ratio reflects the energy dissipation ability of the pile foundation structure during vibration and has an important impact on the stability of the dynamic response. The damping characteristic data provides key energy loss information for the model, enabling it to more accurately evaluate the vibration attenuation of the pile foundation structure under dynamic loads and helping to improve the prediction ability of the seismic and wind load resistance performance of the structure. Through the time series analysis of the natural frequency characteristic data, the modal shape data, and the damping characteristic data, the law of the dynamic characteristics of the pile foundation structure changing with time is evaluated to obtain the vibration modal characteristic data. The time-varying characteristic evaluation can capture the gradual changes of the structural characteristics over time, especially the performance degradation caused by loads, environmental impacts, or material aging. This step provides the attenuation characteristics and time-varying response of the pile foundation structure under long-term dynamic loads, provides a basis for the long-term prediction and reliability assessment of the model, helps the model dynamically adjust the prediction parameters, and improves the adaptability of the prediction. Generally speaking, these steps provide an in-depth analysis of the dynamic characteristics of the pile foundation structure, enabling the model to more comprehensively understand and predict the response changes of the structure under long-term and various dynamic working conditions. By extracting the natural frequency, modal shape, and damping characteristics and conducting time-varying analysis on them, the prediction accuracy and reliability of the model are greatly improved, which can effectively support the durability and safety management of the pile foundation structure.

[0077] Preferably, step S2 includes the following steps:

[0078] Step S21: Extract the multi-source load characteristics from the pile foundation structure condition data to obtain the multi-source load characteristic data;

[0079] In the embodiment of the present invention for extracting the multi-source load characteristics of the pile foundation structure, firstly, the monitored environmental loads are analyzed for sub-item characteristics, including wave loads, wind loads, seismic loads, etc. The amplitude, frequency and acting direction of each load are extracted through Fourier analysis. For example, the seismic wave and wave impact data obtained through acceleration sensors and pressure sensors are used to obtain the average amplitude of the wave load as 5 kN / m through feature extraction 2 , the frequency is 0.6 Hz, and the acting direction is perpendicular to the shoreline. The multi-source load characteristic data provides an accurate data input basis for the subsequent coupling effect analysis.

[0080] Step S22: Based on the multi-source load characteristic data, perform load coupling effect analysis, and perform coupling calculation through the mutual influence and superposition effect between environmental loads to generate coupling load data;

[0081] In the embodiment of the present invention for load coupling effect analysis, by superposing the actions of wave loads and wind loads, finite element software is used for coupling calculation. Through the mutual influence between environmental loads, such as the amplification effect of wave impact on wind pressure, coupling load data is generated. Specifically, in the analysis, it is assumed that the wave load and the wind load are orthogonally superposed, and the calculation results show that the superposition effect makes the peak value of the coupling load reach 7.5 kN / m 2 , and the influence of wind pressure on the pile foundation structure under different wind speeds is considered. The coupling load data reflects the load distribution under complex environmental conditions.

[0082] Step S23: Analyze the spatio-temporal distribution characteristics, action effects and variation laws of the multi-source load characteristic data and the coupling load data to obtain load action data;

[0083] The embodiment of the present invention analyzes the spatio-temporal distribution characteristics of the multi-source load characteristic data and the coupling load data to identify the concentrated areas and variation laws of the load actions. Statistical methods are used to analyze the time series data and combined with spatial grid division. For example, in the area from 0 to 2 m at the bottom of the pile foundation, the load action frequency is high and the amplitude is significant. Through analysis, the time-averaged load value in this area is 6 kN / m 2 , and the peak value is 8 kN / m 2 . This load action data can reflect the force mode of the pile foundation under the change of environmental loads and provide support for the structural safety assessment.

[0084] Step S24: Simulate the response of the pile foundation under different load combination conditions according to the coupling load data, and extract the soil-pile interaction characteristics to generate soil-pile acting force data, where the response simulation under different load combination conditions includes soil resistance, pile body deformation and soil-pile interface contact characteristics;

[0085] In an embodiment of the present invention, the response of a pile foundation under different load combinations is simulated based on coupled load data, and the characteristics of the soil-pile interaction under different conditions are analyzed using the finite element method. Through simulation, it is found that when the wind speed increases to 12 m / s, the lateral displacement of the pile foundation increases, the soil resistance increases to 10 kN / m, and the maximum deformation of the pile body is 3 mm. The contact characteristics of the soil-pile interface show that there is a strong restraint effect on the soil at a depth of 2 m. The analysis results reveal the non-linear response of the soil-pile system under load combinations, and the generated soil-pile interaction force data provides necessary parameter support for the structural stability analysis.

[0086] Step S25: Generate the stress distribution, deformation coordination, and constraint conditions between the soil and the pile foundation based on the soil-pile interaction force data and the load action data, so as to obtain the soil-pile interaction characteristic data;

[0087] In an embodiment of the present invention, based on the soil-pile interaction force data and combined with the load action data, the stress distribution and constraint conditions between the pile foundation and the surrounding soil are evaluated. Through stress field analysis, it is found that the stress concentration area generated by the pile foundation under wave load is located at the bottom of the pile foundation, the pressure value reaches 15 MPa, and the soil constraint is mainly around the pile body at a depth of 1 m to 3 m. The soil-pile interaction characteristic data provides a basis for the design and durability analysis of the pile foundation structure by accurately simulating the binding force of the soil on the pile foundation under load.

[0088] Step S26: Conduct an analysis of soil-pile energy dissipation and stress path evolution based on the soil-pile interaction force data and the soil-pile interaction characteristic data, so as to obtain the soil-pile response data.

[0089] In an embodiment of the present invention, the soil-pile interaction force data and the soil-pile interaction characteristic data are used to conduct energy dissipation and stress path analysis. By calculating the stress paths under different load conditions, it is found that when under strong wave load, the dissipated energy of the soil-pile system increases significantly, reaching 50 J / m 2 , and the stress path shows a non-linear change characteristic. This analysis reveals the energy dissipation mechanism of the pile foundation structure under continuous load and the change trend of the internal stress path. The generated soil-pile response data can be used to predict the performance degradation of the pile foundation under long-term loading.

[0090] The present invention extracts the characteristics of different load sources (such as wind, waves, earthquakes, etc.) in the working condition data of the pile foundation structure, and obtains multi-source load characteristic data. This step helps to decompose the individual action characteristics of various loads, providing data support for the subsequent coupling effect analysis. Extracting multi-source load characteristic data enables the model to distinguish the independent influences of different loads, thus providing detailed information for load characteristic analysis and laying a foundation for accurately predicting the behavior of the pile foundation under combined loads. Based on the multi-source load characteristic data, the load coupling effect analysis is carried out, and combined with the mutual influence between environmental loads, the coupled load data is generated. The coupling effect analysis reveals the superposition effect and mutual influence between multi-source loads, especially the comprehensive effect in the cases of wave and wind loads, earthquake and soil reaction, etc. This analysis provides accurate load input for load combination and the response of the pile foundation under complex working conditions, making the model more realistic and reliable when simulating actual environmental conditions and improving the accuracy of load prediction. Through the analysis of multi-source load characteristic data and coupled load data, the spatio-temporal distribution characteristics of load action, as well as the action effect and variation law, are obtained. This step provides an important reference for understanding the distribution characteristics of loads in the spatial range and time dimension, especially the application characteristics of multi-source loads at different positions and different times. This analysis provides detailed dynamic input for the overall response of the pile foundation under load action, helping to improve the spatial accuracy and time consistency of prediction, and especially being able to better capture the dynamic changes of loads in a complex environment. Based on the coupled load data, the responses of the pile foundation under different load combinations are simulated, and the soil-pile interaction characteristics such as soil resistance, pile body deformation, and soil-pile interface contact characteristics are extracted to generate soil-pile interaction force data. This simulation covers the responses of the pile foundation under different working condition combinations and deeply depicts the complex characteristics of soil-pile interaction. This step provides real soil-pile interface characteristic input for the model, enhancing the model's ability to predict the dynamic response of the pile foundation structure under different load conditions, especially helping to identify and estimate the potential failure modes of the soil-pile system. Based on the soil-pile interaction force data and load action data, the stress distribution, deformation coordination, and constraint conditions between the soil and the pile foundation are analyzed to generate soil-pile interaction characteristic data. The analysis of stress distribution and constraint conditions provides the internal force distribution of the pile foundation under external load action, further revealing the stress transfer and mutual coordination characteristics between the soil and the pile. By generating soil-pile interaction characteristic data, the model can accurately reflect the stress state and constraint conditions of the pile foundation under different load working conditions, improving the prediction of the stability of the pile foundation structure under complex load conditions and the evaluation of its load-bearing capacity. Based on the soil-pile interaction force data and soil-pile interaction characteristic data, the energy dissipation and stress path evolution analysis of the soil-pile system are carried out to obtain soil-pile response data. This step evaluates the dynamic changes of energy dissipation and stress transfer path at the soil-pile interface under load action, providing the response behavior of the pile foundation under complex stress conditions.Energy dissipation and stress path analysis reveal the attenuation characteristics and stress evolution laws of the soil-pile system under long-term and variable loads, providing a scientific basis for the durability and anti-fatigue ability of pile foundations. The model can thereby optimize the safety prediction of pile foundation structures during long-term operation and effectively evaluate potential failure risks. In summary, these steps, through detailed load characteristic analysis, coupled effect calculation, generation of soil-pile interaction characteristics, and energy dissipation analysis, enable the model to have high accuracy and reliability in predicting pile foundation structures under multi-source load conditions. These analyses significantly improve the prediction of the load adaptability and dynamic stability of pile foundation structures, providing scientific support for the long-term reliability and durability management of pile foundations in complex environments.

[0091] Preferably, step S3 includes the following steps:

[0092] Step S31: Conduct a material mechanics performance analysis on the soil-pile response data to obtain material performance characteristic data, where the material mechanics performance analysis includes material strength degradation assessment, elastic modulus change analysis, plastic deformation characteristic extraction, and fatigue damage degree calculation;

[0093] In the material mechanics performance analysis of the soil-pile response data in the embodiment of the present invention, by fitting the stress-strain curves of the pile foundation under different loads, the strength degradation of the pile foundation material is evaluated, and the change trend of the elastic modulus is analyzed. In specific operations, the residual strain data after each loading is extracted for plastic deformation assessment, and the fatigue damage degree is calculated in combination with the measured data. For example, the elastic modulus of the concrete material in the top area of the pile foundation decreases by about 10% within 10 years, and the fatigue damage accumulation ratio reaches 15%. The strength degradation rate of the material increases significantly under impact loads. The generated material performance characteristic data provides an effective reference for the durability analysis of the pile foundation.

[0094] Step S32: Conduct a material performance deterioration assessment based on the material performance characteristic data and the long-term monitoring data of the environment where the pile foundation is located to obtain material state data;

[0095] In the embodiment of the present invention, the material performance characteristic data and the long-term monitoring data of the environment where the pile foundation is located are combined to conduct a deterioration assessment of the pile foundation material performance. Through the historical temperature and humidity data and the degree of salt spray erosion, the deterioration state of the material is quantitatively analyzed, and the deterioration rate is calculated. In practical applications, assume that a pile foundation in a coastal area is long-term exposed to high humidity and high salt environments. The deterioration assessment shows that the strength of the surface concrete of the pile foundation decreases by about 15%, and the number of internal microcracks increases. The generated material state data is used to formulate subsequent maintenance and repair plans.

[0096] Step S33: Use environmental monitoring equipment to collect environmental parameters such as temperature, humidity, and pH value to obtain environmental parameter data;

[0097] In the embodiment of the present invention, environmental monitoring equipment is used to collect environmental parameters such as temperature, humidity, and pH value in the area around the pile foundation in real time, and environmental parameter data is generated. The specific method is to use distributed temperature and humidity sensors and pH sensors to collect data every 1 hour. For example, the average temperature around the pile foundation is detected to be 20°C, the humidity is 85%, and the pH value is 6.5. These environmental parameter data reflect the changes in the environmental conditions around the pile foundation and provide basic data for the subsequent evaluation of the acceleration degradation factor.

[0098] Step S34: Evaluate the acceleration degradation factor based on the environmental parameter data and the material state data, so as to obtain environmental impact characteristic data;

[0099] In the embodiment of the present invention, the environmental parameter data and the material state data are used to evaluate the acceleration degradation factor, and the acceleration degradation effect of the environment on the pile foundation material is calculated. By establishing an acceleration degradation factor model, the influence of parameters such as humidity, temperature, and pH value on the material degradation rate is quantified. In a certain corrosion environment, when the humidity increases by 10%, the material degradation speed increases by about 5%; when the temperature rises to 30°C, the degradation speed increases to 1.5 times the original. The generated environmental impact characteristic data provides important support for the durability analysis of the pile foundation material under different environmental conditions.

[0100] Step S35: Extract time-domain features and frequency-domain features from the real-time monitoring parameter data, so as to obtain multi-dimensional monitoring feature data;

[0101] In the embodiment of the present invention, time-domain and frequency-domain features are extracted from the real-time monitoring parameter data to obtain the force and displacement characteristics of the pile foundation at different time periods. The fast Fourier transform is used to perform spectrum analysis on the data to identify the main frequency distribution and time-domain characteristics of the displacement and load. For example, the main vibration frequency of the pile foundation structure is extracted as 1.2 Hz, and the displacement peak reaches 3 mm during the typhoon. The multi-dimensional monitoring feature data can reflect the dynamic response characteristics of the pile foundation in a complex environment and provide rich input data for the subsequent construction of the correlation model.

[0102] Step S36: Establish a displacement-environment-load correlation model based on the multi-dimensional monitoring feature data and the environmental impact characteristic data, and extract features from the correlation model to obtain correlation feature data.

[0103] In the embodiments of the present invention, an association model of displacement - environment - load is established based on multi - dimensional monitoring feature data and environmental impact feature data to analyze the displacement response relationship of the pile foundation structure under different environmental and load conditions. By machine learning algorithms such as random forest, key features in the association model are extracted, such as the significant variation law of displacement under wind load and high - humidity conditions. Calculations show that when the humidity increases to more than 90% during a typhoon, the displacement peak increases to 1.8 times that of normal times. The generated associated feature data provides a crucial analysis basis for the long - term stability assessment of the pile foundation structure.

[0104] The present invention analyzes the soil-pile response data for material strength degradation, elastic modulus change, plastic deformation characteristics, and fatigue damage degree to obtain material property characteristic data. This analysis can reveal the internal mechanical property changes of the pile foundation under the action of load and environmental stress, especially the reduction of the strength and elasticity of the pile foundation material. These mechanical characteristic analyses provide key parameters for the assessment of the dynamic stability and durability of the material, enabling the model to accurately reflect the bearing capacity of the material in long-term prediction and improving the fatigue life assessment ability of the pile foundation. By combining the material property characteristic data with the long-term monitoring data of the environment where the pile foundation is located, the degradation state of the pile foundation material can be evaluated to generate material state data. This step reflects the performance changes of the pile foundation material under the long-term environmental influence, especially the degradation rate and influencing factors. The material degradation assessment provides a scientific basis for the durability management and safety prediction of the pile foundation structure, enabling the model to more accurately evaluate the service life of the pile foundation under complex working conditions according to different environmental factors and effectively preventing structural failures caused by material degradation. Using environmental monitoring equipment to collect environmental parameter data such as temperature, humidity, and pH value reflects the actual environmental conditions around the pile foundation. Factors such as temperature, humidity, and acidity directly affect the degradation rate of the pile foundation material, especially in harsh environments such as high humidity and high acidity. Collecting environmental parameter data provides real and detailed data support for subsequent degradation assessment and environmental impact analysis, helping to identify key environmental factors that have a significant impact on material degradation and making the degradation assessment more accurate. Based on the environmental parameter data and the material state data, through the analysis of the acceleration degradation factor, the acceleration degradation effects of environmental factors such as temperature, humidity, and acidity on the pile foundation material can be identified to generate environmental impact characteristic data. The acceleration degradation factor assessment provides the specific influence weights of different environmental conditions on the material performance. The acceleration degradation factor analysis helps in the long-term degradation prediction of the pile foundation in harsh environments, especially in the durability management under extreme climate conditions, provides a scientific basis for the pile foundation maintenance cycle and material selection, and improves the environmental adaptability of the structure. Extracting the time-domain and frequency-domain characteristics of the real-time monitoring parameter data to obtain multi-dimensional monitoring characteristic data, including the time series and frequency characteristics of displacement, stress, and load changes. These multi-dimensional characteristic data provide comprehensive monitoring information for the dynamic response of the pile foundation and can effectively reflect the behavior patterns of the pile foundation under various loads. The combined analysis of the time-domain and frequency-domain characteristics improves the response characteristic recognition ability of the model, enabling the prediction model to dynamically track the structural state changes and ensuring the real-time monitoring and accurate prediction of the pile foundation behavior. Establishing a displacement-environment-load correlation model based on the multi-dimensional monitoring characteristic data and the environmental impact characteristic data, and generating correlation characteristic data by extracting the characteristics in this model. This model reflects the mutual relationship among the pile foundation displacement, environmental factors, and load action, and is the core support for model prediction.The construction and feature extraction of the correlation model achieve the prediction of the overall behavior of pile foundations under complex environmental conditions, enabling the model to dynamically adjust prediction parameters, improving the response accuracy and prediction applicability under various working conditions. Generally speaking, these steps enhance the adaptability and accuracy of the model to long-term deterioration and variable environments through the detailed analysis of material properties and environmental factors, the extraction of environmental impact factors, and the construction of the displacement-environment-load correlation model. This process ensures the safety of pile foundations during long-term use, providing accurate and reliable data support and a scientific basis for the maintenance management and performance prediction of pile foundations in complex environments.

[0105] Preferably, step S34 includes the following steps:

[0106] Evaluate the acceleration deterioration factor based on the environmental parameter data and the material state data to obtain the environmental impact characteristic data, where the acceleration deterioration factor evaluation is specifically the temperature stress coefficient of the Arrhenius formula, the humidity expansion coefficient is evaluated through the Arrhenius humidity-temperature acceleration model, and the corrosion rate coefficient is calculated based on Faraday's law.

[0107] In the embodiment of the present invention, when evaluating the acceleration deterioration factor, first use the temperature and humidity values in the environmental parameter data to calculate the temperature stress coefficient and the humidity expansion coefficient. In specific operations, select the monitoring data of a certain pile foundation, record the average temperature as 25°C and the humidity as 80%. According to the Arrhenius formula, calculate the temperature stress coefficient K T The formula for where A is the pre-factor, E a is the activation energy, R is the gas constant, and T is the absolute temperature. Assume the activation energy E a is 50 kJ / mol, and obtain K T ≈1.5. Then, apply the Arrhenius humidity-temperature acceleration model to calculate the humidity expansion coefficient K H , for example, if the environmental humidity varies between 70% and 90%, set the expansion coefficient to 0.02% / °C, and assume the expansion acceleration factor at the current humidity level is 1.3, and obtain K H ≈0.026. Through the above calculations, the combined data of the temperature stress coefficient and the humidity expansion coefficient are obtained, forming preliminary environmental impact characteristic data, laying a foundation for further corrosion rate evaluation. When evaluating the corrosion rate coefficient, calculate using Faraday's law, and the specific formula is where I is the current, n is the number of transfers, F is the Faraday constant, and dM / dt is the corrosion rate. Assume that the current density in a certain corrosion environment is obtained through electrochemical measurement as 10 mA / cm 2 , estimate the number of transfers n as 2, and calculate the corrosion rate coefficient as where F is taken as approximately 96485 C / mol, and obtain KC ≈0.00012 g / cm 2 ·h. By this method, combined with the previous environmental stress and humidity expansion data, through comprehensive analysis, the finally generated environmental impact characteristic data provides a more comprehensive assessment for the deterioration of pile foundation materials and a scientific basis for the formulation of subsequent structural monitoring and maintenance strategies.

[0108] During the long-term use of pile foundations, environmental factors such as temperature, humidity, and corrosion can significantly affect the deterioration rate of materials. Therefore, the evaluation of acceleration deterioration factors such as temperature, humidity, and corrosion can provide a key basis for the long-term reliability prediction of pile foundation materials. Scientifically evaluating the acceleration deterioration factors using environmental parameter data and material state data, the generated environmental impact characteristic data is of great significance for the stability prediction and maintenance strategy optimization of pile foundation structures. Calculating the temperature stress coefficient through the Arrhenius formula can accurately evaluate the impact of temperature on the material deterioration rate. The Arrhenius formula provides a reliable scientific model for predicting the deterioration process of materials under different temperature conditions by describing the rate of chemical reactions with temperature changes. The evaluation of the temperature stress coefficient can help predict the service life of pile foundations under extreme temperatures or long-term temperature fluctuations, especially identifying the trend of accelerated deterioration under high-temperature conditions, facilitating the formulation of protection and maintenance strategies in high-temperature environments and extending the durability of pile foundation materials. The humidity expansion coefficient is evaluated through the Arrhenius humidity-temperature acceleration model, which can reflect the impact of humidity on the expansion and deterioration of pile foundation materials at different temperatures. The humidity-temperature acceleration model comprehensively considers the coupling effect of humidity and temperature, making the evaluation of humidity impact more accurate, especially valuable for humidity-sensitive materials. The evaluation of the humidity expansion coefficient can reveal the expansion and deterioration speed of pile foundation materials in high-humidity environments, facilitating the pretreatment of pile foundation materials or the selection of more suitable materials in high-humidity environments to prevent material property degradation caused by humidity, such as strength reduction or durability decline. Calculating the corrosion rate coefficient through Faraday's law, based on the relationship between the electrochemical reaction rate and current density of materials, quantifies the impact of corrosion on material deterioration. Faraday's law can relatively accurately describe the weight loss rate of materials under different corrosion conditions, providing a scientific basis for deterioration prediction in long-term immersion and salt spray environments. The evaluation of the corrosion rate coefficient is particularly applicable to the pile foundation application scenarios in coastal or high-salt areas, which can predict the failure trend of pile foundation materials in corrosive environments, help select materials with higher corrosion resistance, or formulate anti-corrosion construction and maintenance measures in advance. Through the evaluation of the temperature stress coefficient, humidity expansion coefficient, and corrosion rate coefficient, the generated environmental impact characteristic data provides a multi-factor coupling characteristic system for the material deterioration prediction of pile foundation structures. This comprehensive environmental impact analysis model can accurately reflect the deterioration mode of materials in complex environments. The acquisition of environmental impact characteristic data enhances the life prediction model of pile foundation materials, enabling the model to effectively distinguish the deterioration impacts of different environmental conditions and improving the accuracy of structural health monitoring. At the same time, considering multiple acceleration deterioration factors makes the model applicable to more diverse environments, ensuring the wide applicability and scientific nature of the prediction. In summary, this step, based on the evaluation of acceleration deterioration factors of different environmental parameters, provides a quantitative evaluation criterion for the long-term deterioration prediction of pile foundation materials.By evaluating the coefficients of key environmental factors such as temperature, humidity, and corrosion, it can help the model more accurately reflect the reliability of the pile foundation under extreme conditions, guide maintenance and material selection, extend the service life of the structure, and improve safety and economy.

[0109] Preferably, step S4 includes the following steps:

[0110] Step S41: Construct a deep neural network structure based on the associated feature data, and use the backpropagation algorithm for model training to obtain the prediction model parameters;

[0111] In the embodiment of the present invention, a deep neural network structure is constructed based on the associated feature data. A network model suitable for time series prediction, such as a long short-term memory (LSTM) network, is selected as the basic architecture. Assume that the input layer contains 50 neurons, then two LSTM layers are set, each layer contains 100 units, and finally a fully connected layer is connected to output the displacement prediction value of the pile foundation. Then, using the collected historical data, the model is trained using the backpropagation algorithm. During the specific training process, the learning rate is set to 0.001, the mean squared error (MSE) is used as the loss function, the model parameters are adjusted through each epoch, and 1000 epochs are continuously performed until the loss value converges to a preset threshold (such as 0.01). After the training is completed, the model parameters are saved for subsequent prediction use.

[0112] Step S42: Perform time series decomposition on the load action data, and extract the periodic characteristics, trend characteristics, and random characteristics of the load to obtain the load time series characteristic model;

[0113] In the embodiment of the present invention, time series decomposition is performed on the load action data. First, a seasonal decomposition method (such as STL decomposition) is used to decompose the load data into trend, seasonal, and random components. Assume that the load data monitored over a period of time (such as 12 months) is [200, 220, 250, 230, 210, 240, 300, 280, 270, 290, 310, 320] kN. First, calculate its moving average to obtain the trend component. Then, calculate the periodic component (such as the monthly fluctuation), and extract the random component, which is analyzed through the statsmodels library in Python. Finally, a load time series characteristic model is formed to describe the main trend, periodic fluctuation, and random characteristics of the load change.

[0114] Step S43: Perform coupling analysis on the load time series characteristic model and the prediction model parameters to construct a time series correlation prediction model, thereby obtaining the time series prediction parameter data;

[0115] When the embodiment of the present invention performs coupled analysis on the load time-series feature model and the prediction model parameters, first, the results of load time-series feature extraction are combined with the parameters of the trained deep neural network. By using the extracted trend and seasonal components as inputs, the constructed prediction model is used for the next coupled analysis. Suppose the extracted trend features are [215, 230, 260] kN, and the seasonal features are [20, -10, 30] kN. Combining historical load data for modeling, in a weighted sum manner, the load time-series feature model is coupled with the deep learning model to form a time-series correlation prediction model, and time-series prediction parameter data is obtained to provide the predicted load change of the pile foundation within a future period of time.

[0116] Step S44: Optimize the prediction accuracy based on the attention mechanism according to the time-series prediction parameter data, so as to generate dynamic prediction control data;

[0117] The embodiment of the present invention optimizes the prediction accuracy based on the attention mechanism according to the time-series prediction parameter data. Specifically, when implemented, an attention layer is first introduced into the deep neural network to better capture the input features that have the greatest impact on future predictions. Suppose through training, it is found that the load change has a greater impact on the overall prediction result during a specific period (such as the peak period of each day). Therefore, higher weights are assigned to these features in the model, and by adjusting the attention mechanism parameters for optimization, the model pays more attention to the load feature data at critical moments, thereby generating dynamic prediction control data and improving the accuracy of the prediction result.

[0118] Step S45: Analyze the prediction error of the dynamic prediction control data, calculate the deviation between the predicted value and the measured value, so as to generate prediction error feature data;

[0119] When the embodiment of the present invention analyzes the prediction error of the dynamic prediction control data, first, the actual monitoring values and the predicted values are collected for comparison. Suppose at a certain time point, the predicted value is 310 kN, and the actual monitoring value is 300 kN. The prediction error is calculated as 310 - 300 = 10 kN. Such calculations are performed for all predicted values and measured values to obtain a set of error data. The root mean square error (RMSE) and mean absolute percentage error (MAPE) are calculated through statistical analysis to evaluate the overall prediction performance of the model and generate prediction error feature data.

[0120] Step S46: Perform sensitivity analysis on the prediction model parameters according to the prediction error feature data, so as to obtain parameter sensitivity data;

[0121] In an embodiment of the present invention, sensitivity analysis of the prediction model parameters is performed based on the prediction error characteristic data. First, a set of reference parameter values (such as learning rate, number of hidden layer units, etc.) are set, and the change of the prediction error is monitored when the parameters change. Using a sensitivity analysis method (such as local sensitivity analysis), the learning rate is changed (for example, adjusted from 0.001 to 0.0005) and prediction is performed again, and the degree of change of the prediction error is observed. By recording the degree of influence of different parameter changes on the prediction result, parameter sensitivity data is obtained, and the parameter that has the greatest influence on the prediction effect is identified, providing a basis for subsequent optimization.

[0122] Step S47: Adaptively update the prediction model parameters according to the dynamic prediction control data and the parameter sensitivity data, so as to obtain optimized model parameters.

[0123] In an embodiment of the present invention, the prediction model parameters are adaptively updated according to the dynamic prediction control data and the parameter sensitivity data. First, the key parameters (such as learning rate and number of hidden layer units) identified in the sensitivity analysis are selected, and the model is retrained using an adaptive algorithm (such as Adam optimizer). Suppose it is found in the sensitivity analysis that reducing the learning rate can significantly improve the prediction performance, so the learning rate is adjusted to 0.0005, and the model is continuously trained until the error reaches a new lower level. After adaptive update, the optimized model parameters are obtained, further improving the accuracy and reliability of the model in future load prediction.

[0124] The present invention utilizes a deep neural network (DNN) to process multi-source associated feature data of pile foundation structures. The DNN can capture complex displacement feature patterns in a non-linear and high-dimensional feature space. The backpropagation algorithm continuously optimizes the model parameters through error feedback, enabling the model to better adapt to the actual data features. The prediction model parameters obtained through training the deep neural network have high generalization ability and can handle diverse working conditions and complex load characteristics, thereby improving the prediction accuracy of the model. By performing time series decomposition on the load action data, periodic, trend, and random features are extracted, which can accurately describe the dynamic change characteristics of the load. This time series decomposition enables the model to identify different impacts of the load on the pile foundation structure and further construct a load time series feature model. The feature model obtained through time series decomposition can help identify the load change pattern, making the model more forward-looking when predicting future load actions, improving the load prediction accuracy, and providing more stable input data for the displacement prediction of the pile foundation structure. By coupling and analyzing the load time series feature model with the prediction model parameters, a time series correlation prediction model is constructed. This coupling analysis can better correlate the response relationship between the load and the displacement, enabling the prediction model to dynamically reflect the real-time impact of the load on the pile foundation structure. Constructing the time series correlation prediction model can improve the dynamic adaptability of the prediction model, enabling the model to also have good response capabilities when dealing with sudden load changes and enhancing the timeliness prediction ability of the model for pile foundation displacements. By introducing an attention mechanism, the model can focus on the key load features that have a greater impact on the prediction, reducing the interference of irrelevant or secondary features, thereby improving the accuracy of the prediction results. The introduction of the attention mechanism effectively improves the model's analytical ability for complex load data, making the prediction results more robust and accurate, and providing high-quality data support for the displacement prediction of the pile foundation structure. By calculating the deviation between the predicted value and the measured value, prediction error feature data is obtained to identify the error distribution of the model under different working conditions. This error analysis can promptly discover the sources of deviation of the model and provide a basis for subsequent optimization. The error feature data can be an important reference for model correction, helping to identify the factors that may cause prediction deviations under specific working conditions and improving the stability and credibility of the model. By performing sensitivity analysis to evaluate the impact of different prediction model parameters on the prediction results, the key parameters that are most sensitive to the prediction results are identified, thereby optimizing the model structure. Obtaining the parameter sensitivity data can guide the model to more specifically adjust the key parameters during future adaptive update processes, avoiding blind optimization, and improving the precision and convergence of the model. Based on the dynamic prediction control data and the sensitivity analysis results, the model parameter adaptive update ensures that the model can respond to working condition changes in real time, enabling the model to continuously adapt to environmental and load changes, and thus maintaining a high level of prediction accuracy.The optimized model parameters obtained through adaptive update enable the model to dynamically adjust to cope with complex load and environmental factor changes, ensuring the long-term effectiveness and accuracy of displacement prediction results, and contributing to improving the safety and reliability of pile foundation structures during long-term use. The above steps achieve high-precision dynamic prediction of pile foundation displacement through the combination of deep learning and time series analysis, as well as the introduction of attention mechanism and adaptive optimization. This method can effectively capture the impact of load changes on pile foundation structures, provide reliable data support for long-term safety monitoring, risk warning, and maintenance optimization, make the displacement prediction of pile foundation structures more accurate in complex environments, and ensure the long-term stability and safety of the structure.

[0125] Preferably, step S5 includes the following steps:

[0126] Step S51: Construct a lateral displacement prediction model using the optimized model parameters and perform continuous prediction calculations to obtain an initial displacement prediction sequence;

[0127] In the embodiment of the present invention, a lateral displacement prediction model is constructed using the optimized model parameters, and a suitable machine learning algorithm, such as support vector machine (SVM) or recurrent neural network (RNN), is selected for model training based on historical monitoring data (such as the displacement data of the pile foundation in the past year). Assume that the historical data contains 1000 data points, each data point including environmental factors such as temperature, load, and humidity as input features, and an optimized deep neural network is used as the regression model. Continuous prediction calculations are performed using the optimized parameters, a time window (such as the next 30 days) is set to predict the displacement of each day step by step, the results of each prediction are recorded, and finally an initial displacement prediction sequence is obtained. For example, the predicted displacement values may be [0.2, 0.25, 0.3, 0.28, 0.31,...] meters.

[0128] Step S52: Extract time series features based on displacement change trends, periodic features, and mutation features from the initial displacement prediction data, and perform prediction accuracy verification to obtain displacement prediction data;

[0129] When the embodiments of the present invention extract the time series features of the initial data for displacement prediction, first, the change trend of the displacement is calculated, and a linear regression model is used to analyze the trend features of the prediction sequence to obtain the slope of the trend line. For example, if the trend line equation of the displacement sequence is y = 0.02x + 0.1, it means that the displacement increases by 0.02 meters per time unit. Then, the periodic features are extracted, and the fast Fourier transform (FFT) is used to analyze the periodic components of the displacement data. Suppose the monthly periodic feature is obtained as 0.1 meter. Finally, the mutation features are analyzed, and the quantile regression method is used to detect whether there are obvious mutation points. For example, the displacement suddenly increases to 0.5 meters at a certain point in time. These features are comprehensively used to verify the prediction accuracy. By comparing with the actual observed data, the accuracy of the prediction data is ensured, and thus the processed displacement prediction data is obtained.

[0130] Step S53: Quantitatively evaluate the uncertainty of the displacement prediction data to obtain prediction uncertainty data;

[0131] When the embodiments of the present invention quantitatively evaluate the uncertainty of the displacement prediction data, first, the environmental factors related to the displacement and their possible change ranges are collected (for example, the change range of temperature is 5 to 35 °C, and the humidity is 30% to 90%). Through the Monte Carlo simulation method, a large number of random samples are generated. Suppose 10,000 simulations are performed, and the displacement prediction values under each simulation are calculated to obtain a set of distribution results. Then, the variance and standard deviation of these results are analyzed to quantify the uncertainty and determine the uncertainty range. For example, the standard deviation of the predicted displacement is 0.02 meters. Finally, the prediction uncertainty data is obtained to describe the reliability of the model for future displacements.

[0132] Step S54: Calculate the reliability index of the displacement prediction result according to the prediction uncertainty data to obtain prediction reliability data, where the reliability index includes the prediction error distribution, confidence interval, and prediction accuracy evaluation;

[0133] When the embodiments of the present invention calculate the reliability index of the displacement prediction result according to the prediction uncertainty data, first, the prediction error distribution is analyzed, the error between the predicted value and the measured value is calculated and a normal distribution test is performed. Suppose the mean of the obtained error distribution is 0.01 meter and the standard deviation is 0.02 meter. Then, the confidence interval is calculated based on the prediction error distribution. For example, using a 95% confidence interval range of [0.01 - 1.96 * 0.02, 0.01 + 1.96 * 0.02], the obtained confidence interval is [-0.03, 0.05]. Finally, combined with the prediction accuracy evaluation index (such as the root mean square error), these data are summarized into prediction reliability data to comprehensively evaluate the reliability of the prediction result.

[0134] Step S55: Perform correction processing on the displacement prediction data based on the prediction uncertainty data and the prediction reliability data, so as to obtain the corrected prediction result data.

[0135] When the embodiment of the present invention performs correction processing on the displacement prediction data based on the prediction uncertainty data and the prediction reliability data, first, correction parameters are set according to the results of uncertainty analysis. Assume that according to the previous analysis, the correction coefficient is set to 0.8. Multiply the initial displacement prediction value by the correction coefficient, and at the same time add the confidence interval of the prediction error, so that the corrected displacement = initial displacement × 0.8 + error range. For example, for a certain prediction with an initial value of 0.25 meters, after correction calculation, it becomes 0.25 × 0.8 + 0.02 = 0.22 meters. At this time, the corrected prediction result data obtained through correction processing is a more reliable displacement value, so as to improve the overall prediction accuracy and credibility.

[0136] Based on the model parameters optimized previously, the present invention constructs a model capable of continuously predicting the lateral displacement of pile foundations. This process integrates the effects of dynamic loads and environmental factors into the model to generate an initial displacement prediction sequence. Through this continuous prediction calculation, the changes in pile foundation displacement can be tracked in real time, providing a basis for subsequent engineering decisions. At the same time, the initial prediction sequence lays a foundation for further feature extraction and accuracy verification. Analyze the initial displacement prediction data to extract the trend features, periodic features, and mutation features of displacement changes, thereby verifying the prediction accuracy. The goal of this step is to identify the key patterns in the data and confirm the effectiveness of the prediction. The extraction of time series features can provide a deeper understanding of the dynamic behavior of pile foundation displacement, ensuring that the prediction model not only generates data but also provides meaningful trend analysis. In addition, the verification of prediction accuracy improves the credibility of the prediction results, providing a reliable basis for subsequent steps. Through quantitative evaluation, analyze the possible uncertainty factors in the prediction data, including model errors, measurement errors, and environmental changes, etc., to generate prediction uncertainty data. The evaluation of uncertainty quantification provides a more comprehensive understanding of the prediction results, which can reveal the vulnerability and sensitivity of the model under different working conditions, providing important information for risk assessment and decision-making, and enhancing the effectiveness of management and response measures. According to the prediction uncertainty data, calculate reliability indicators including prediction error distribution, confidence interval, and prediction accuracy evaluation, etc., to generate prediction reliability data. These indicators help to identify the stability and credibility of the prediction results. The calculation of reliability indicators can provide a quantitative confidence value for the prediction results, enabling engineers and decision-makers to more clearly understand the reliability of the prediction results, thereby enhancing the trust in the prediction results and facilitating more scientific measures to be taken in practical applications. Based on the prediction uncertainty data and prediction reliability data, perform correction processing on the displacement prediction results to generate corrected prediction result data. This process ensures that the prediction data can better reflect the actual situation. The correction processing makes the final prediction results more accurate and reliable, helping to eliminate potential errors and biases, enabling engineering personnel to rely on higher-quality data support when making decisions and evaluations, and thus improving the safety and sustainability of engineering projects. The entire prediction process not only focuses on the optimization of model parameters and the realization of continuous prediction, but also emphasizes the evaluation and correction of prediction results. Such a process ensures that the pile foundation displacement prediction model has high accuracy and reliability in a dynamic and complex environment, can effectively support engineering management and maintenance decisions, and ensure the safety and stability of pile foundation structures. This systematic processing method provides strong technical support for the long-term monitoring and management of pile foundation projects.

[0137] Therefore, in any aspect, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, it is intended to encompass all changes falling within the meaning and scope of the equivalent elements of the application documents within the present invention.

[0138] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the broadest scope consistent with the principles and novel features invented herein.

Claims

1. A method for constructing a lateral displacement prediction model for pile foundation of a high-pile wharf, characterized in that: The following steps are involved: Step S1: monitor the displacement, stress and load parameters of the pile foundation of the high-pile wharf and generate real-time monitoring parameter data; Identify the working condition of the pile foundation structure through real-time monitoring of parameter data and generate working condition data of the pile foundation structure; Step S2: Perform multi-source load characteristic analysis on the pile foundation structure working condition data, and calculate the load coupling effect to generate load action data; perform soil-pile interaction analysis based on the load action data to generate soil-pile response data; Step S3: evaluating material performance degradation according to soil-pile response data to generate material status data; analyzing the impact of environmental factors based on temperature, humidity and corrosion according to the material status data to generate environmental impact characteristic data; extracting features from real-time monitoring parameter data, and performing dynamic feature fusion based on correlation analysis on environmental correction data to obtain displacement-environment-load correlation characteristic data; Step S4: using the associated feature data to train the deep learning model and obtain the prediction model parameters; performing time series feature analysis on the load action data and the prediction model parameters, and optimizing the prediction accuracy based on the attention mechanism to generate dynamic prediction control data; adaptively updating the prediction model parameters according to the dynamic prediction control data to obtain optimized model parameters; Step S5: Continuously predict the lateral displacement of the pile foundation using the optimized model parameters to generate displacement prediction data; quantify the uncertainty and conduct reliability evaluation on the displacement prediction data to generate corrected prediction result data.

2. The method for constructing a lateral displacement prediction model for pile foundation of a high-pile wharf according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: using multi-source sensors installed on the pile foundation of the high-pile wharf and the soil in the surrounding area to collect sensor parameters in real time, so as to obtain pile foundation displacement data, pile foundation force data and environmental load data, wherein the multi-source sensors include displacement sensors, stress sensors and load sensors; Step S12: pre-processing the pile foundation displacement data, pile foundation force data and environmental load data, and performing unified data dimension processing to obtain real-time monitoring parameter data; Step S13: performing static working condition identification on the real-time monitoring parameter data to obtain static working condition characteristic data, wherein the static working condition identification specifically includes static load working condition classification, structural deformation feature extraction, stress distribution pattern recognition and support condition evaluation; Step S14: performing dynamic response feature extraction, vibration modal analysis, dynamic load feature identification, and dynamic operating condition identification for structural dynamic characteristic evaluation based on the real-time monitoring parameter data, thereby obtaining dynamic operating condition feature data; Step S15: Perform information collaborative fusion processing on the static working condition characteristic data and the dynamic working condition characteristic data, so as to obtain the pile foundation structure working condition data.

3. The method for constructing a lateral displacement prediction model for pile foundation of a high-pile wharf according to claim 2, characterized in that: Step S13 includes the following steps: Step S131: Classifying the environmental load data in the real-time monitoring parameter data based on the loading mode, loading amplitude and loading sequence, and identifying the load condition according to the preset static load condition classification standard, thereby obtaining static load condition classification data; Step S132: extracting deformation characteristics of the pile foundation displacement data, and calculating characteristic quantities of displacement amplitude, deformation trend, deformation rate and deformation mode, thereby obtaining structural deformation characteristic data; Step S133: performing stress distribution pattern recognition based on stress distribution law, stress concentration area, stress transfer path and stress level based on the pile foundation stress data, thereby obtaining stress distribution pattern data; Step S134: evaluating the interaction relationship between the pile foundation and the soil according to the pile foundation displacement data and the pile foundation force data, and generating constraint conditions for the pile end constraint, soil constraint and pile body constraint of the pile foundation, thereby obtaining support condition data; Step S135: feature fusion of static load condition classification data, structural deformation characteristic data, stress distribution pattern data and support condition data to obtain static condition characteristic data.

4. The method for constructing a lateral displacement prediction model for pile foundation of a high-pile wharf according to claim 3, characterized in that: Step S14 includes the following steps: Step S141: extracting dynamic response characteristics of dynamic response parameters of the pile foundation structure based on displacement time history, acceleration response and stress fluctuation based on the real-time monitoring parameter data, thereby obtaining dynamic response characteristic data; Step S142: performing vibration modal analysis according to the dynamic response characteristic data, thereby obtaining vibration modal characteristic data; Step S143: extracting dynamic load characteristics of wave load, wind load and earthquake load from environmental load data, and performing parameter statistics of amplitude, frequency, phase and action direction of dynamic load characteristics, so as to obtain dynamic load characteristic data; Step S144: evaluating dynamic characteristic parameters according to the dynamic response characteristic data and the vibration modal characteristic data to obtain structural dynamic characteristic data, wherein the dynamic characteristic parameters include stiffness characteristics, mass distribution and energy dissipation of the pile foundation structure; Step S145: performing characteristic correlation analysis on the dynamic response characteristic data, the vibration mode characteristic data, the dynamic load characteristic data and the structural dynamic characteristic data, so as to obtain dynamic working condition characteristic data.

5. The method for constructing a lateral displacement prediction model for pile foundation of a high-pile wharf according to claim 4, characterized in that: Step S142 includes the following steps: Step S1421: extracting the natural frequency of the pile foundation structure from the dynamic response characteristic data to obtain natural frequency characteristic data; Step S1422: performing modal vibration shape analysis based on the natural frequency characteristic data, extracting the vibration modal shape of the pile foundation structure, and calculating the amplitude and phase characteristics of each order mode, thereby obtaining modal vibration shape data; Step S1423: Calculating the damping ratio of each mode of the pile foundation structure based on the half-power method for the dynamic response characteristic data and the modal vibration shape data, thereby obtaining damping characteristic data; Step S1424: Analyze the changing rules of the natural frequency characteristic data, the modal vibration shape data and the damping characteristic data based on the time series, and evaluate the time-varying characteristics of the dynamic characteristics of the pile foundation structure, so as to obtain the vibration modal characteristic data.

6. The method for constructing a lateral displacement prediction model for pile foundation of a high-pile wharf according to claim 5, characterized in that: Step S2 includes the following steps: Step S21: extracting multi-source load characteristics from the pile foundation structure working condition data, thereby obtaining multi-source load characteristic data; Step S22: performing load coupling effect analysis based on multi-source load characteristic data, and performing coupling calculation through the mutual influence and superposition effect between environmental loads to generate coupling load data; Step S23: Analyze the temporal and spatial distribution characteristics, effects and change rules of the load action on the multi-source load characteristic data and the coupled load data, so as to obtain the load action data; Step S24: simulating the response of the pile foundation under different load combination conditions according to the coupled load data, extracting the soil-pile interaction characteristics, and generating soil-pile force data, wherein the response simulation under different load combination conditions includes soil resistance, pile deformation, and soil-pile interface contact characteristics; Step S25: generating stress distribution, deformation coordination and constraint conditions between the soil and the pile foundation based on the soil-pile force data and the load action data, thereby obtaining soil-pile interaction characteristic data; Step S26: Perform soil-pile energy dissipation and stress path evolution analysis based on the soil-pile force data and the soil-pile interaction characteristic data, so as to obtain soil-pile response data.

7. The method for constructing a lateral displacement prediction model for pile foundation of a high-pile wharf according to claim 6, characterized in that: Step S3 includes the following steps: Step S31: performing material mechanical property analysis on the soil-pile response data to obtain material performance characteristic data, wherein the material mechanical property analysis includes material strength degradation assessment, elastic modulus variation analysis, plastic deformation feature extraction, and fatigue damage degree calculation; Step S32: evaluating material performance degradation based on the material performance characteristic data and the pre-acquired long-term monitoring data of the environment in which the pile foundation is located, thereby obtaining material status data; Step S33: using environmental monitoring equipment to collect environmental parameters such as temperature, humidity and pH value, thereby obtaining environmental parameter data; Step S34: evaluating the accelerated degradation factor according to the environmental parameter data and the material status data, thereby obtaining environmental impact characteristic data; Step S35: extracting time domain features and frequency domain features from the real-time monitoring parameter data, thereby obtaining multi-dimensional monitoring feature data; Step S36: establishing a displacement-environment-load association model based on the multi-dimensional monitoring characteristic data and the environmental impact characteristic data, and performing feature extraction on the association model to obtain association characteristic data.

8. The method for constructing a lateral displacement prediction model for pile foundation of a high-pile wharf according to claim 7, characterized in that: Step S34 includes the following steps: The accelerated degradation factor is evaluated based on the environmental parameter data and material status data to obtain the environmental impact characteristic data. The accelerated degradation factor is evaluated specifically by the temperature stress coefficient through the Arrhenius formula, the humidity expansion coefficient is evaluated through the Arrhenius humidity-temperature acceleration model, and the corrosion rate coefficient is calculated based on Faraday's law.

9. The method for constructing a lateral displacement prediction model for pile foundation of a high-pile wharf according to claim 8, characterized in that: Step S4 includes the following steps: Step S41: constructing a deep neural network structure based on the associated feature data, and using a back propagation algorithm to perform model training, thereby obtaining prediction model parameters; Step S42: performing time series decomposition on the load action data, and extracting the periodicity characteristics, trend characteristics and random characteristics of the load, so as to obtain a load time series characteristic model; Step S43: performing coupling analysis on the load time series characteristic model and the prediction model parameters to construct a time series correlation prediction model, thereby obtaining time series prediction parameter data; Step S44: optimizing the prediction accuracy based on the attention mechanism according to the time series prediction parameter data, thereby generating dynamic prediction control data; Step S45: performing prediction error analysis on the dynamic prediction control data, calculating the deviation between the predicted value and the measured value, thereby generating prediction error characteristic data; Step S46: performing sensitivity analysis on the prediction model parameters according to the prediction error characteristic data, thereby obtaining parameter sensitivity data; Step S47: Adaptively update the prediction model parameters according to the dynamic prediction control data and the parameter sensitivity data, so as to obtain optimized model parameters.

10. The method for constructing a lateral displacement prediction model for pile foundation of a high-pile wharf according to claim 9, characterized in that: Step S5 includes the following steps: Step S51: constructing a lateral displacement prediction model using optimized model parameters, and performing continuous prediction calculations to obtain an initial displacement prediction sequence; Step S52: extracting time series features based on displacement change trend, periodic features and mutation features from the initial displacement prediction data, and verifying the prediction accuracy, thereby obtaining displacement prediction data; Step S53: performing uncertainty quantitative evaluation on the displacement prediction data, thereby obtaining prediction uncertainty data; Step S54: Calculating the reliability index of the displacement prediction result according to the prediction uncertainty data, thereby obtaining prediction reliability data, wherein the reliability index includes prediction error distribution, confidence interval and prediction accuracy evaluation; Step S55: Correct the displacement prediction data based on the prediction uncertainty data and the prediction reliability data to obtain corrected prediction result data.

Images