A pile soft soil deformation stress cooperative early warning method, system, device and medium

Abstract

The present application provides a kind of pile soft soil deformation stress coordination early warning method, system, equipment and medium, it is related to underground engineering technical field, the monitoring data of soft soil around pile is obtained based on monitoring point, the monitoring point includes pile body monitoring point and soil monitoring point;Feature extraction is carried out to monitoring data, and monitoring index reflecting the evolution of deformation and stress of soft soil around pile is obtained based on information contribution degree difference and sensitivity selection;Model construction is carried out based on monitoring data and monitoring index, and by considering risk assessment and spatial heterogeneity, the deformation-stress coordination atlas of soft soil around pile is obtained;Based on the deformation-stress coordination atlas, the anomaly of soft soil around pile is identified, and the abnormal area is obtained by embedding learning and clustering analysis, and hierarchical early warning is carried out according to the abnormal area, to obtain the early warning result of soft soil around pile.The present application solves the problem that the existing early warning method of soft soil around pile lacks dynamic perception and joint early warning of pile-soil interaction process.

Description

Technical Field

[0001] This invention relates to the field of underground engineering technology, and more specifically, to a method, system, equipment, and medium for coordinated early warning of deformation stress in soft soil around piles. Background Technology

[0002] In underground engineering, pile foundations are often used in soft soil areas to enhance the bearing capacity and stability of the foundation. However, under the long-term operation of infrastructure such as railways or under extreme loads (such as train loads, rainstorms, earthquakes, etc.), the soft soil around the piles is prone to deformation accumulation and strength reduction, which can lead to pile tilting, sinking, or even structural instability.

[0003] Current railway health monitoring primarily focuses on the structure itself (e.g., pile strain, displacement) or overall soil settlement, but lacks dynamic perception and joint early warning mechanisms for pile-soil interaction processes. Traditional methods have the following limitations: they only monitor single indicators (stress or deformation) in the pile or soil, failing to reflect the coupling relationship between soft soil deformation and stress changes in real time. Furthermore, the deformation and stress of the soft soil around the pile exhibit significant spatial heterogeneity (e.g., vertical settlement is transmitted layer by layer, and horizontal stress attenuates radially). However, existing technologies typically treat monitoring points as independent units, failing to construct mechanical correlation models between nodes, making it difficult to identify the spatial diffusion patterns of local anomalies. Additionally, existing methods often issue early warnings based on the exceeding limits of indicators at a single monitoring point, making it difficult to correlate discrete anomalies into physically meaningful continuous regions, hindering engineers from quickly locating risk sources.

[0004] Therefore, there is an urgent need for a collaborative early warning method that considers the coupling relationship between soft soil deformation and stress change, as well as the spatial heterogeneity of deformation and stress in the soft soil around the pile, to achieve dynamic perception and joint early warning of the pile-soil interaction process and obtain accurate early warning results for the soft soil around the pile. Summary of the Invention

[0005] The purpose of this invention is to provide a method, system, and device for coordinated early warning of deformation stress in soft soil around piles, in order to improve the aforementioned problems. To achieve the above objective, the technical solution adopted by this invention is as follows:

[0006] Firstly, this application provides a method for coordinated early warning of deformation stress in soft soil around piles, including:

[0007] Monitoring data of the soft soil around the pile is obtained based on monitoring points. The monitoring data includes pile deformation data, pile stress data, soil deformation data, soil stress data and environmental data. The monitoring points include pile monitoring points and soil monitoring points.

[0008] Feature extraction was performed on the monitoring data, and monitoring indicators reflecting the deformation and stress evolution of the soft soil around the pile were obtained based on the differences in information contribution and sensitivity selection.

[0009] Based on monitoring data and indicators, a model was constructed, and by considering risk assessment and spatial heterogeneity, a deformation-stress synergistic spectrum of the soft soil around the pile was obtained.

[0010] Anomaly identification of soft soil around piles is performed based on deformation-stress synergistic maps. Anomaly regions are obtained through embedding learning and cluster analysis, and graded early warnings are given based on the anomaly regions to obtain early warning results for soft soil around piles.

[0011] Secondly, this application also provides a collaborative early warning system for deformation stress in soft soil around piles, including:

[0012] The acquisition unit is used to acquire monitoring data of the soft soil around the pile based on the monitoring points. The monitoring data includes pile deformation data, pile stress data, soil deformation data, soil stress data and environmental data. The monitoring points include pile monitoring points and soil monitoring points.

[0013] The feature extraction unit is used to extract features from the monitoring data. Based on the differences in information contribution and sensitivity selection, monitoring indicators reflecting the deformation and stress evolution of the soft soil around the pile are obtained.

[0014] The building unit is used to build a model based on monitoring data and monitoring indicators. By taking into account risk assessment and spatial heterogeneity, the deformation-stress synergy spectrum of the soft soil around the pile is obtained.

[0015] The early warning unit is used to identify anomalies in soft soil around piles based on deformation-stress synergistic maps. It obtains abnormal areas through embedding learning and cluster analysis, and performs hierarchical early warnings based on the abnormal areas to obtain early warning results for soft soil around piles.

[0016] Thirdly, this application also provides a collaborative early warning device for deformation stress of soft soil around piles, including:

[0017] Memory, used to store computer programs;

[0018] A processor is used to implement the steps of the collaborative early warning method for deformation stress of soft soil around piles when executing the computer program.

[0019] Fourthly, this application also provides a readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described method for coordinated early warning of deformation stress in soft soil around piles.

[0020] The beneficial effects of this invention are as follows: By preprocessing and extracting features from multi-source monitoring data of soft soil around piles, this invention selects monitoring indicators reflecting soil deformation and stress evolution, achieving dynamic quantification and effective tracking of the soil state around the pile. Combining predicted mechanical response values ​​with monitoring data, a comprehensive risk factor input space is constructed, enhancing the model's depth and generalization ability, laying the foundation for risk assessment. Based on engineering structural logic, a deformation-stress synergistic map is constructed, and graph neural networks are used for embedding learning to capture the spatial synergistic characteristics of soil deformation and stress. Through node anomaly calculation and cluster analysis, accurate identification and graded early warning of abnormal areas are achieved. This invention enhances the accuracy of anomaly identification and the practicality of early warning, providing a scientific and real-time risk early warning method for the safety monitoring of soft soil around piles, improving engineering safety assurance and management efficiency, and possessing significant application and promotion value.

[0021] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a schematic diagram of the collaborative early warning method for deformation stress of soft soil around piles as described in an embodiment of the present invention;

[0024] Figure 2 This is a schematic diagram of the pile and soil as described in an embodiment of the present invention;

[0025] Figure 3 This is a schematic diagram of the structure of the pile perimeter soft soil deformation stress collaborative early warning system described in this embodiment of the invention;

[0026] Figure 4 This is a schematic diagram of the structure of the pile perimeter soft soil deformation stress collaborative early warning device described in this embodiment of the invention.

[0027] The diagram is labeled as follows: 901, acquisition unit; 902, feature extraction unit; 903, construction unit; 904, early warning unit; 800, pile-surround soft soil deformation stress collaborative early warning device; 801, processor; 802, memory; 803, multimedia component; 804, I / O interface; 805, communication component. Detailed Implementation

[0028] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0029] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0030] Example 1:

[0031] This embodiment provides a method for coordinated early warning of deformation stress in soft soil around piles.

[0032] See Figure 1 The figure shows that the method includes steps S1, S2, S3, and S4.

[0033] Step S1: Obtain monitoring data of the soft soil around the pile based on the monitoring points. The monitoring data includes pile deformation data, pile stress data, soil deformation data, soil stress data and environmental data. The monitoring points include pile monitoring points and soil monitoring points.

[0034] In this embodiment, multiple monitoring points are selected for early warning of the soft soil area surrounding the pile. This is due to the spatial heterogeneity of the soft soil area and the localized nature of the pile-soil interaction. Soft soil mechanical parameters (such as compression modulus, shear strength, and void ratio) often exhibit significant spatial variability. Even within the same location, soil strength, moisture content, and settlement rate may differ at different points. Single-point data cannot reflect these spatial differences. Furthermore, the stress transfer and deformation effects of the pile foundation on the surrounding soft soil are locally attenuated. Therefore, single-point monitoring is insufficient to reflect the overall health status of the pile-soil system.

[0035] Step S1 includes:

[0036] Step S11: Set the monitoring area for the pile foundation and arrange multiple monitoring points based on the monitoring area;

[0037] In this embodiment, based on engineering requirements and the distribution of the pile foundation structure, a monitoring range is defined in the pile foundation and the surrounding soft soil area. According to the geological conditions, pile foundation type, and load distribution characteristics of the monitoring area, pile body monitoring points and soil monitoring points are rationally deployed to achieve spatially distributed monitoring of the pile-soil system. The deployment of monitoring points should cover different depths and horizontal positions to ensure the representativeness and completeness of the monitoring data.

[0038] Step S12: Collect pile deformation data and pile stress data based on pile monitoring points. The pile deformation data includes pile horizontal displacement, pile vertical displacement and pile strain. The pile stress data includes pile axial stress and pile shear stress.

[0039] In this embodiment, monitoring sensors (such as fiber optic sensors, strain gauges, displacement gauges, etc.) installed on the inner surface of the pile body are used to collect deformation and stress information generated by the pile body during operation. Among them, the horizontal displacement of the pile body reflects the lateral stress deformation, the vertical displacement of the pile body reflects the vertical settlement, the pile body strain reflects the location of local stress concentration or damage, the axial stress of the pile body characterizes the longitudinal load transfer, and the shear stress of the pile body reflects the relationship between lateral sliding and soil interaction.

[0040] Step S13: Collect soil deformation data and soil stress data of soft soil around the pile based on soil monitoring points. The soil deformation data includes vertical settlement and horizontal displacement of soft soil around the pile, and the soil stress data includes soil pressure, pore water pressure and soil strain.

[0041] In this embodiment, the vertical settlement and horizontal displacement of the soft soil around the pile reflect the overall subsidence trend of the foundation and the deformation expansion of the soft soil around the pile in the lateral direction, respectively.

[0042] Step S14: Obtain environmental data based on monitoring points, including temperature and groundwater level.

[0043] In this embodiment, environmental factors have a significant impact on the mechanical properties of soft soil. For example, temperature affects the physical state and material properties of the soil, while groundwater level affects pore pressure changes, effective stress state, and soil strength. In particular, environmental factors play a dominant role in the response of soft soil during heavy rain or seasonal changes. Therefore, environmental data needs to be collected simultaneously for subsequent model correction and anomaly interpretation.

[0044] This step also preprocesses the collected pile deformation data, pile stress data, soil deformation data, soil stress data, and environmental data to obtain monitoring data. Specifically, since different sensors (such as pile strain gauges, earth pressure sensors, and groundwater level gauges) may have different sampling frequencies or time lags, all data need to be aligned to a unified time reference. Methods such as interpolation, resampling, or sliding window averaging are used to ensure that various types of data are comparable at the same time point. At the same time, due to the irregular spatial distribution, their spatial coordinates need to be standardized to construct a coordinate index, or mapped to a standard reference system based on a geographic / engineering coordinate system (such as station number + depth) so that a deformation-stress co-location map can be established later.

[0045] like Figure 2 The diagram shown is a schematic of a pile and soil system. Figure 2 The data provides soil layer properties; soil layers at different depths possess different characteristics, therefore the collected data will vary. Figure 2 The pile shown is a conventional straight pile, 4000cm long and 160cm in diameter. Eight cross-sections are set on the pile surface, corresponding to 20 stress gauges, for collecting pile stress data. Simultaneously, corresponding pile strain gauges are also installed on the pile surface to collect pile deformation data, and soil pressure sensors, groundwater level gauges, and other sensors are placed in the soil layer.

[0046] Simultaneously, data cleaning and repair are performed to remove or fill missing values ​​caused by equipment failure or communication interruption, correct or remove abnormal abrupt changes (such as jumps caused by sensor drift), and sliding median filtering or Z-score detection can be used. At the same time, physical consistency verification is performed on all data (such as settlement cannot be negative and pore pressure should be within a reasonable range).

[0047] Because different data dimensions have different dimensions, their numerical ranges and units vary significantly. If the raw data is used directly for subsequent correlation analysis or scoring calculations, indicators with larger dimensions will have a dominant influence on the calculation process, such as the Pearson correlation coefficient and weighted scoring, leading to an imbalance in indicator selection. Therefore, the processed data is normalized (using Min-Max normalization or Z-score standardization).

[0048] After the above steps, the collected pile deformation data, pile stress data, soil deformation data, soil stress data and environmental data are integrated into a unified data structure in time and space to obtain monitoring data.

[0049] Step S2: Extract features from the monitoring data, and obtain monitoring indicators reflecting the deformation and stress evolution of the soft soil around the pile based on the differences in information contribution and sensitivity selection.

[0050] In this embodiment, the acquired monitoring data is high-dimensional, redundant, and heterogeneous, with differences in information redundancy and contribution among the data. Therefore, relevant indicators are extracted. Different indicators contribute differently to reflecting the pile-soil interaction state. Therefore, feature selection is used to select the set of monitoring indicators that are most sensitive to and representative of the deformation and stress coupling evolution.

[0051] In step S2, the steps for obtaining the monitoring indicators are as follows:

[0052] Step S21: Construct a first candidate indicator using monitoring data, wherein the first candidate indicator includes multiple candidate feature indicators;

[0053] In this embodiment, the first candidate index is a potential sensitive index that reflects the coupled deformation-stress-environmental evolution process of the soft soil area around the pile, which is obtained through preprocessed soil deformation data, soil stress data and environmental data.

[0054] Candidate feature indicators based on soil deformation data include the cumulative vertical settlement, cumulative horizontal displacement, daily settlement increment, displacement change rate, coefficient of variation of settlement / horizontal displacement change, deformation ratio at different depths (such as surface / deep settlement ratio), and the ratio of surface horizontal displacement to pile top displacement.

[0055] Candidate feature indices based on soil stress data include soil pore water pressure, pore pressure change rate, vertical effective stress, horizontal effective stress, effective stress ratio, stress path offset (initial-current), and stress change fluctuation amplitude (standard deviation).

[0056] Candidate feature indicators based on environmental data include soil temperature variation, groundwater level variation, daily rainfall, and settlement variation within one day after rainfall.

[0057] It also includes candidate feature indicators based on multiple data structures, including the correlation coefficient between soil displacement and pore pressure, water level-pore pressure ratio, pore pressure recovery coefficient (after drainage or rainfall), settlement-stress delay time, multi-point settlement synergy coefficient, and the angle between the main displacement direction and the line connecting the pile center.

[0058] Multiple extracted candidate feature indicators were used as the first candidate indicators, which covered the deformation response characteristics of soft soil, the evolution of stress and pore pressure, the influence of environmental disturbance, and the multimodal coupling behavior pattern.

[0059] Step S22: Calculate the Pearson correlation coefficient matrix between multiple candidate feature indicators using the first candidate indicator, and obtain the Pearson correlation coefficient between every two candidate feature indicators;

[0060] In this embodiment, multiple candidate feature indicators are extracted from monitoring data such as soil settlement, stress, pore pressure, displacement, temperature and water level. Although these indicators can reflect information from different perspectives, there are highly correlated redundant indicators. If used directly, it will lead to information weight shift and cover up the truly key influencing factors. Therefore, it is necessary to use the Pearson correlation coefficient matrix to evaluate the linear correlation between indicators and remove highly redundant indicators to improve the efficiency and accuracy of subsequent analysis.

[0061] In this step, the n candidate feature indicators are each grouped into a multidimensional vector, forming an m×n indicator matrix, where m represents the number of samples. The Pearson correlation coefficient is calculated for any two columns of the indicator matrix to obtain the Pearson correlation coefficient between any two candidate feature indicators.

[0062] Step S23: Redundancy removal of multiple candidate feature indicators is performed using the Pearson correlation coefficient to obtain the second candidate indicator;

[0063] In this embodiment, in order to remove highly correlated features and avoid information redundancy or weight shift in subsequent weighting analysis, a redundancy discrimination threshold of 0.85 is set. When the Pearson correlation coefficient of any two candidate indicators is greater than the redundancy discrimination threshold, they are considered redundant.

[0064] After identifying all indicator pairs exceeding the redundancy threshold, a redundant pair set is constructed. For each redundant pair in the redundant pair set, the average correlation coefficient of each candidate feature indicator is calculated. Indicators with lower correlation are retained first. If multiple indicators are redundant and have no obvious advantages or disadvantages, selection can be based on information entropy, magnitude of change, domain knowledge, or stability in subsequent analysis, finally yielding the second candidate indicator.

[0065] Step S24: Calculate the comprehensive score of each candidate feature indicator in the second candidate indicator based on the subjective and objective weighting fusion method. After sorting the second candidate indicators based on the comprehensive scores, select a preset number of candidate feature indicators in the second candidate indicator as monitoring indicators.

[0066] In this embodiment, the ability of different indicators to reflect the deformation and stress evolution of soft soil around piles is influenced by their own data change characteristics (such as information entropy, volatility, etc.) as well as by subjective knowledge such as engineering experience, expert judgment, and physical meaning. For example, the vertical settlement rate of soil is usually a direct indicator of soft soil deformation and has a high weight of engineering experience, while the pore pressure fluctuation frequency may be important in some areas but not significant in others.

[0067] Therefore, both objective and subjective weighting methods have inherent biases. To achieve a scientific assessment of the importance of indicators, a combined objective and subjective weighting mechanism is introduced, namely the entropy weighting method (objective) + the AHP method (subjective). This achieves a balance between data-driven and knowledge-driven approaches, improving the credibility and engineering adaptability of the final selected indicators, resulting in the final monitoring indicators chosen.

[0068] In this step, the objective weight of each candidate feature indicator in the second candidate indicator is first calculated based on the entropy weight method. Specifically, the information entropy of each candidate feature indicator is calculated, and then the objective weight of each candidate feature indicator is calculated based on the information entropy.

[0069] Then, the subjective weights of each candidate feature indicator in the second candidate indicator are calculated using the Analytic Hierarchy Process (AHP). Specifically, an indicator pair comparison matrix is ​​constructed based on the experience of experts or technicians, where each element represents the relative importance between two candidate feature indicators. The weights of the indicator pair comparison matrix are then calculated using the eigenvector method to obtain the subjective weight matrix of the candidate feature indicators. Since human assignment introduces subjectivity, a consistency check is required to determine the reasonableness of the subjective weight matrix. Therefore, the consistency ratio of the subjective weight matrix is ​​calculated and a consistency check is performed. If the consistency ratio is less than 0.1, the indicator pair comparison matrix needs to be adjusted and recalculated until the consistency check is satisfied, resulting in the final subjective weight matrix.

[0070] The subjective weights of each candidate feature indicator are obtained through a subjective weight matrix. The combined weights of the subjective and objective weights are then used to calculate the comprehensive score for each candidate feature indicator in the second candidate indicator set. Based on these comprehensive scores, the top 6 candidate feature indicators are selected as the final monitoring indicators.

[0071] Step S3: Based on monitoring data and indicators, a model is constructed. By considering risk assessment and spatial heterogeneity, the deformation-stress synergistic spectrum of the soft soil around the pile is obtained.

[0072] Step S3 includes:

[0073] Step S31: Construct a mechanical response prediction model based on monitoring data that reflects the deformation-stress coupling relationship of the soft soil around the pile;

[0074] Understandably, since traditional methods cannot explain why settlement continues to develop slowly after construction and why risk evolution is not detected under combined working conditions, this step uses a mechanical response prediction model to model and couple pore pressure, stress diffusion and time to explain the delayed settlement mechanism. At the same time, through model fitting, short-term and medium-term response prediction curves are provided to construct virtual disturbance scenarios and identify potential risk areas in advance.

[0075] In step S31, the construction steps of the mechanical response prediction model are as follows:

[0076] Step S311: Construct a basic mechanical response model based on the consolidation and seepage-mechanical coupling theory and multi-field coupling mechanism. The basic mechanical response model includes a pore pressure-settlement response model, a pile-soil contact interface strain function, a force-seepage-temperature coupling control equation set, and a stress evolution and settlement response function.

[0077] In this embodiment, the pore pressure-settlement response model is as follows:

[0078]

[0079] In the formula, u represents the pore water pressure, and c u This represents the consolidation coefficient, and the pore pressure diffusion capacity. Denotes the divergence operator, t represents the pore pressure gradient and t represents time.

[0080] The strain function at the pile-soil interface is as follows:

[0081] ∈ p-s (t)=f(Δu(t),Δσ h (t),T(t))

[0082] In the formula, ∈ p-s (t) represents the strain at the pile-soil interface at time t, Δu(t) represents the change in pore water pressure at time t, and Δσ h (t) represents the change in horizontal stress at time t, T(t) represents the soil temperature at time t, f(Δu(t), Δσ) h (t),T(t)) represent the expressions with respect to Δu(t), Δσ h Functions of T(t) and T(t).

[0083] The force-permeability-temperature coupling governing equations are simplified to two-dimensional plane strain equations, including stress equilibrium equations, constitutive relations, and permeability governing equations, which are as follows:

[0084]

[0085] σ=D:(∈-∈ p -∈ T )

[0086]

[0087] In the formula, Let σ denote the divergence operator, g denote the stress tensor, ρ denote the gravitational acceleration vector, D denote the constitutive stiffness matrix, and ∈ denote the total strain tensor. pRepresents plastic strain, ∈ T The value represents thermal expansion strain, k represents the permeability coefficient, μ represents the dynamic viscosity of the liquid, and ∈ v : indicates volumetric strain, and : indicates double contraction.

[0088] Based on empirical construction of stress evolution and settlement response functions, specifically:

[0089] S(t) = S0 + a·log(1 + b·Δσ) v (t))+c·u(t)

[0090] In the formula, S(t) represents the settlement of the soft soil around the pile at time t, S0 represents the initial settlement value, a, b, and c represent the fitting coefficients, and Δσ v u(t) represents the vertical stress increment at time t, and u(t) represents the pore water pressure at time t.

[0091] Step S312: Process the monitoring data to obtain a structured monitoring data matrix;

[0092] Step S313: Fit the mechanical response basic model based on the structured monitoring data matrix to obtain a mechanical prediction response model that reflects the deformation-stress coupling relationship of the soft soil around the pile.

[0093] Step S32: Obtain the predicted mechanical response value of each soil monitoring point at future time using the mechanical response prediction model;

[0094] Step S33: Input the predicted mechanical response value and monitoring index into the risk assessment model to conduct risk assessment and obtain the assessment results of each soil monitoring point in the soft soil around the pile. The assessment results include risk score, status label and evolution trend.

[0095] Understandably, since a single monitoring indicator cannot comprehensively measure engineering risks and there is a lack of a unified quantitative model to comprehensively analyze and classify monitoring indicators and risks, this step involves constructing a multi-indicator-driven risk assessment model to map complex and multi-dimensional monitoring information into an interpretable risk score, thereby enabling status identification, trend assessment, and subsequent risk grading and early warning triggering.

[0096] In step S33, the steps for constructing the risk assessment model are as follows:

[0097] Step S331: Construct a risk factor vector using the predicted mechanical response and monitoring indicators, and set each component in the risk factor vector as an input node of the Bayesian network;

[0098] In this embodiment, each monitoring index and each predicted mechanical response value (settlement prediction value, stress prediction value) are used as input nodes of the Bayesian network to establish a complete risk input feature space. This enables the model to not only rely on observation data but also integrate physical prediction information, thereby enhancing the modeling depth and generalization ability and providing a clear and quantifiable input vector for Bayesian network inference.

[0099] Step S332: Set the risk score, state label, and evolution trend as the target node of the Bayesian network;

[0100] In this embodiment, the risk score is a continuous variable (0-100) reflecting the overall risk level, the state label is a categorical variable representing normal, abnormal, and dangerous states, and the evolution trend is a trend judgment variable, such as rising, stable, or falling. The evolution trend refers to the changing trend of the risk score, such as whether the risk score is continuously rising, remaining stable, or gradually decreasing. Therefore, the multi-objective node format supports rich reasoning capabilities, which helps to meet multi-level engineering needs (such as automatic early warning, manual intervention, and long-term trend prediction).

[0101] Step S333: Set up the Bayesian network structure based on the causal relationship between the input nodes and the target nodes;

[0102] In this embodiment, the input node is used as the parent node, the output variable is used as the child node, and the Yesian network structure is established by guiding expert knowledge.

[0103] Step S334: Define the prior distribution of the input nodes and the conditional probability table of the target nodes;

[0104] In this embodiment, the Bayesian network primarily uses discrete variables. The input and target nodes of the continuous variables must first be discretized, divided into different levels based on their value ranges. For example, the risk score can be discretized into levels of 10 points each. Status labels map the risk scores; for instance, the first three levels are labeled as normal, the middle four as abnormal, and the last three as dangerous. The parent node contains the pore pressure change rate, and thresholds are set based on practical engineering experience to divide the corresponding intervals, thus achieving discretization.

[0105] After discretization, the prior distribution of the input nodes is obtained by statistical analysis of historical data. For target nodes with parent nodes, the probability distribution under each combination of parent nodes needs to be defined. This is achieved by calculating the conditional probability table of the target node based on the probability statistics of each state combination in the historical data.

[0106] Step S335: Train a Bayesian network using historical data, update the conditional probability table, and adjust the Bayesian network structure to obtain a risk assessment model.

[0107] In this embodiment, a Bayesian network is trained using historical data (historical response values, historical monitoring indicators, actual risk levels, status labels, trend evolution information, etc.), the conditional probability table is updated, and the Bayesian network structure is fine-tuned to obtain a risk assessment model.

[0108] Step S34: Using the evaluation results as node features and combining the spatial distribution relationship between soil monitoring points, construct the deformation-stress synergy map of the soft soil around the pile.

[0109] In this embodiment, in order to organically integrate the spatial distribution structure with the evaluation results, graph neural network modeling, temporal reasoning and spatial synergistic analysis are performed to obtain the deformation-stress synergistic spectrum of the soft soil around the pile.

[0110] In step S34, constructing the deformation-stress synergistic map of the soft soil around the pile includes:

[0111] Step S341: Define each soil monitoring point as a graph node, and use the evaluation result of each soil monitoring point as the node feature of the corresponding graph node;

[0112] Step S342: Define the edge connection relationship between graph nodes based on the spatial distribution relationship between soil monitoring points;

[0113] In this embodiment, since there is a significant spatial interlayer and directional correlation between soil deformation and stress evolution (vertical settlement usually propagates between upper and lower layers, while horizontal stress is transmitted radially or circumferentially), the edge connection relationship between the nodes in the diagram is constructed using engineering structural logical connections.

[0114] Specifically, the three-dimensional coordinates of each soil monitoring point are obtained. If the horizontal distance between two soil monitoring points is within a small threshold (the horizontal and vertical coordinates are basically the same), the vertical distance is the layer thickness and is set as a vertical edge to simulate the mechanical transmission and response coupling between the upper and lower soil layers.

[0115] If two soil monitoring points are located at the same depth (the difference in vertical coordinates is less than the preset depth threshold) and the same radius (the horizontal distance from the pile center is less than half the preset threshold), and they are angularly distributed around the pile, then connect them and set up circumferential edges to simulate the coordinated evolution of stress and settlement in different directions in the same soil layer.

[0116] Meanwhile, when two soil monitoring points are located at the same depth and angle, the soil monitoring points located at the same angle but different radii are connected to obtain a radial edge, forming a path from the center of the pile outward. This simulates the process of the pile transmitting mechanical influence to the surrounding soil and captures the deformation and stress evolution law of the pile load propagating radially through the soil.

[0117] By connecting upper and lower layers, connecting the same layer, and connecting the pile and soil in the near and far fields, the final adjacency matrix is ​​obtained, which is the edge connection relationship between graph nodes.

[0118] Step S343: Construct a deformation-stress synergistic map of the soft soil around the pile based on graph nodes, node features, and edge connection relationships.

[0119] Step S4: Based on the deformation-stress synergistic spectrum, anomaly identification of soft soil around piles is performed. Anomaly areas are obtained through embedding learning and cluster analysis, and graded early warnings are given according to the anomaly areas to obtain the early warning results of soft soil around piles.

[0120] In step S4, obtaining the early warning result for the soft soil around the pile includes:

[0121] Step S41: Based on graph neural network, perform embedding learning on deformation-stress co-location graph to obtain a low-dimensional embedding representation of each graph node;

[0122] In this embodiment, a graph neural network (using a GCN network in this step) is designed and trained. Through a multi-layer message passing mechanism, neighbor node information is aggregated, and low-dimensional embedding representations of nodes are learned. The low-dimensional embedding representations capture the complex spatial relationships and feature coupling information between graph nodes and are mapped to a semantically rich feature space.

[0123] Step S42: Calculate the Euclidean distance between each graph node based on the low-dimensional embedding representation, and calculate the outlier of each graph node using the Euclidean distance;

[0124] In this embodiment, when obtaining the neighboring nodes of each graph node, the number of neighboring nodes is dynamically adjusted. In sparse node areas (such as far-field soil), the number of neighboring nodes is increased to avoid fluctuations in the mean due to too few neighbors. In dense node areas (such as the area near the pile), the number of neighboring nodes is reduced, focusing on core associated nodes. The area is automatically divided by a density clustering algorithm (such as the DBSCAN algorithm), and an adaptive number of neighboring nodes is allocated to different areas.

[0125] For each graph node and its neighboring nodes, the Euclidean distance of the low-dimensional embedding representation is calculated to generate a node-neighbor distance matrix. Then, based on the node-neighbor distance matrix, the average of all Euclidean distances for each graph node is calculated as the outlier degree of the graph node. Since it essentially calculates the average degree of difference of each graph node in its local neighborhood, a larger average Euclidean distance indicates that the graph node is more outlier.

[0126] Step S43: Perform cluster analysis based on anomaly degree to delineate potential anomalous areas in the soft soil around the piles;

[0127] In this embodiment, when setting the anomaly threshold based on historical data, a moving average method is used to adapt to the time changes in soil condition. For example, the anomaly threshold is calculated using the average value over a preset time period and adjustment parameters. Each graph node is marked according to its anomaly degree; if it exceeds the anomaly threshold, it is considered an anomaly node.

[0128] In order to organize discrete anomalous nodes into continuous spatial regions to facilitate engineering identification and location of anomalous regions, clustering algorithms are used to group anomalous nodes and identify spatially adjacent node clusters with high anomalousness. These node clusters are defined as anomalous regions.

[0129] Step S44: Calculate the outlier value of the outlier region by using the outlier degree of all graph nodes within the outlier region;

[0130] In this embodiment, the anomaly degree of all nodes within each anomaly region is statistically calculated (average value is calculated) to obtain the anomaly value of that anomaly region. This allows for the numerical quantification of the overall risk level of the anomaly region, facilitating hierarchical judgment, while reducing the noise impact of single-point anomalies.

[0131] Step S45: Compare the outlier values ​​with the preset graded early warning standards to obtain the early warning results for the soft soil around the pile.

[0132] In this embodiment, a graded early warning threshold (normal, minor warning, moderate warning, severe warning) is set for the abnormal value. The abnormal value in the abnormal area is compared with the graded early warning threshold to determine its warning level. The overall early warning result of the soft soil around the pile is output, including the location of the abnormal area and the corresponding warning level.

[0133] In summary, this invention constructs monitoring indicators reflecting the deformation and stress evolution of soft soil around piles through preprocessing and feature extraction of multi-source monitoring data, achieving accurate quantification and dynamic tracking of the soil state around the pile. A screening strategy combining subjective and objective weighting is employed to eliminate redundant features, effectively improving the representativeness and stability of the indicators and ensuring the scientific rigor and reliability of subsequent analyses. Furthermore, based on the fusion of mechanical response prediction and monitoring data, a comprehensive risk factor input space is constructed, enhancing the model's depth and generalization ability, laying a solid foundation for subsequent risk assessment.

[0134] Furthermore, by combining a graph construction method based on engineering structural logic, the spatial relationships between soil monitoring points are accurately depicted. Low-dimensional embedding learning using graph neural networks fully captures the co-evolution characteristics of soil deformation and stress. Cluster analysis based on node anomaly degrees enables spatial identification and hierarchical early warning of anomalous areas, providing a scientific, accurate, and real-time risk warning method for monitoring the condition of soft soil around piles. This invention not only improves the accuracy of anomaly identification but also enhances the interpretability of early warning results, demonstrating promising prospects for engineering application.

[0135] Example 2:

[0136] like Figure 3 As shown in the figure, this embodiment provides a collaborative early warning system for deformation stress in soft soil around piles. The system includes:

[0137] The acquisition unit 901 is used to acquire monitoring data of the soft soil around the pile based on the monitoring points. The monitoring data includes pile deformation data, pile stress data, soil deformation data, soil stress data and environmental data. The monitoring points include pile monitoring points and soil monitoring points.

[0138] The feature extraction unit 902 is used to extract features from the monitoring data and obtain monitoring indicators reflecting the deformation and stress evolution of the soft soil around the pile based on the differences in information contribution and sensitivity selection.

[0139] Building unit 903 is used to build a model based on monitoring data and monitoring indicators. By considering risk assessment and spatial heterogeneity, the deformation-stress synergy spectrum of the soft soil around the pile is obtained.

[0140] The early warning unit 904 is used to identify anomalies in the soft soil around piles based on the deformation-stress synergistic spectrum. It obtains the abnormal areas through embedding learning and cluster analysis, and performs hierarchical early warning based on the abnormal areas to obtain the early warning results for the soft soil around piles.

[0141] It should be noted that the specific methods by which each module performs operations in the system described in the above embodiments have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0142] Example 3:

[0143] Corresponding to the above method embodiments, this embodiment also provides a pile-surround soft soil deformation stress collaborative early warning device. The pile-surround soft soil deformation stress collaborative early warning device described below and the pile-surround soft soil deformation stress collaborative early warning method described above can be referred to each other.

[0144] Figure 4 This is a block diagram illustrating a pile-surround soft soil deformation stress collaborative early warning device 800 according to an exemplary embodiment. Figure 4 As shown, the pile-surround soft soil deformation stress collaborative early warning device 800 may include: a processor 801 and a memory 802. The pile-surround soft soil deformation stress collaborative early warning device 800 may also include one or more of the following: a multimedia component 803, an I / O interface 804, and a communication component 805.

[0145] The processor 801 controls the overall operation of the pile-surround soft soil deformation stress collaborative early warning device 800 to complete all or part of the steps in the aforementioned pile-surround soft soil deformation stress collaborative early warning method. The memory 802 stores various types of data to support the operation of the pile-surround soft soil deformation stress collaborative early warning device 800. This data may include, for example, instructions for any application or method operating on the pile-surround soft soil deformation stress collaborative early warning device 800, as well as application-related data such as contact data, sent and received messages, images, audio, video, etc. The memory 802 can be implemented using any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The multimedia component 803 may include a screen and an audio component. The screen may be, for example, a touchscreen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signals may be further stored in the memory 802 or transmitted via the communication component 805. The audio component also includes at least one speaker for outputting audio signals. I / O interface 804 provides an interface between processor 801 and other interface modules, such as keyboards, mice, and buttons. These buttons can be virtual or physical. Communication component 805 is used for wired or wireless communication between the pile-surround soft soil deformation stress collaborative early warning device 800 and other devices. Wireless communication includes Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or a combination thereof. Therefore, the corresponding communication component 805 may include a Wi-Fi module, a Bluetooth module, or an NFC module.

[0146] In an exemplary embodiment, the pile-surround soft soil deformation stress collaborative early warning device 800 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described pile-surround soft soil deformation stress collaborative early warning method.

[0147] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, which, when executed by a processor, implements the steps of the above-described method for coordinated early warning of deformation stress in soft soil around piles. For example, the computer-readable storage medium may be the memory 802 including the program instructions, which may be executed by the processor 801 of the device 800 for coordinated early warning of deformation stress in soft soil around piles to complete the above-described method for coordinated early warning of deformation stress in soft soil around piles.

[0148] Example 4:

[0149] Corresponding to the above method embodiments, this embodiment also provides a readable storage medium. The readable storage medium described below can be referred to in conjunction with the above-described method for coordinated early warning of deformation stress in soft soil around piles.

[0150] A readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the pile perimeter soft soil deformation stress collaborative early warning method of the above method embodiments.

[0151] Specifically, the readable storage medium can be a USB flash drive, a portable hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, or any other readable storage medium capable of storing program code.

[0152] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0153] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A pile soft soil deformation stress cooperative early warning method, characterized in that, include: Monitoring data of the soft soil around the pile is obtained based on monitoring points. The monitoring data includes pile deformation data, pile stress data, soil deformation data, soil stress data and environmental data. The monitoring points include pile monitoring points and soil monitoring points. Feature extraction was performed on the monitoring data, and monitoring indicators reflecting the deformation and stress evolution of the soft soil around the pile were obtained based on the differences in information contribution and sensitivity selection. Based on monitoring data and indicators, a model was constructed. By considering risk assessment and spatial heterogeneity, a deformation-stress synergistic spectrum of the soft soil around the pile was obtained, including: A mechanical response prediction model reflecting the deformation-stress coupling relationship of soft soil around piles was constructed based on monitoring data. The mechanical response prediction model is used to obtain the predicted mechanical response value of each soil monitoring point at a future time. The predicted mechanical response and monitoring indicators are input into the risk assessment model for risk assessment, and the assessment results of each soil monitoring point in the soft soil around the pile are obtained. The assessment results include risk score, status label and evolution trend. Using the assessment results as node characteristics and combining the spatial distribution relationship between soil monitoring points, a deformation-stress synergistic map of the soft soil around the pile is constructed. Anomaly identification of soft soil around piles is performed based on deformation-stress synergistic maps. Anomaly regions are obtained through embedding learning and cluster analysis, and graded early warnings are given based on the anomaly regions to obtain early warning results for soft soil around piles.

2. The method for coordinated early warning of deformation stress in soft soil around piles according to claim 1, characterized in that... The steps for obtaining the monitoring indicators are as follows: A first candidate indicator is constructed based on monitoring data, and the first candidate indicator includes multiple candidate feature indicators. The Pearson correlation coefficient matrix between multiple candidate feature indicators is calculated using the first candidate indicator, and the Pearson correlation coefficient between each pair of candidate feature indicators is obtained. Redundancy was eliminated from multiple candidate feature indicators using the Pearson correlation coefficient to obtain the second candidate indicator; The comprehensive score of each candidate feature indicator in the second candidate indicator is calculated based on the subjective and objective weighting fusion method. After the second candidate indicators are ranked based on the comprehensive scores, a preset number of candidate feature indicators in the second candidate indicators are selected as monitoring indicators.

3. The method for coordinated early warning of deformation stress in soft soil around piles according to claim 1, characterized in that... The construction steps of the mechanical response prediction model are as follows: Based on the theory of consolidation and seepage-mechanical coupling and the multi-field coupling mechanism, a basic mechanical response model is constructed. The basic mechanical response model includes a pore pressure-settlement response model, a pile-soil contact interface strain function, a force-seepage-temperature coupling control equation set, and a stress evolution and settlement response function. The monitoring data is processed to obtain a structured monitoring data matrix; By fitting the basic mechanical response model with the structured monitoring data matrix, a mechanical prediction response model reflecting the deformation-stress coupling relationship of the soft soil around the pile is obtained.

4. The method for coordinated early warning of deformation stress in soft soil around piles according to claim 1, characterized in that... The steps for constructing the risk assessment model are as follows: A risk factor vector is constructed using predicted mechanical response values ​​and monitoring indicators, and each component in the risk factor vector is set as an input node of a Bayesian network. Risk score, state label, and evolution trend are set as the target nodes of the Bayesian network; The Bayesian network structure is set based on the causal relationship between the input nodes and the target nodes; Define the prior distribution of the input nodes and the conditional probability table of the target nodes; A risk assessment model is obtained by training a Bayesian network using historical data, updating the conditional probability table, and adjusting the Bayesian network structure.

5. The method for coordinated early warning of deformation stress in soft soil around piles according to claim 1, characterized in that... The deformation-stress synergistic map of the soft soil surrounding the pile includes: Each soil monitoring point is defined as a graph node, and the evaluation result of each soil monitoring point is used as the node feature of the corresponding graph node. Define the edge connection relationship between graph nodes based on the spatial distribution relationship between soil monitoring points; Deformation-stress synergistic graph of soft soil around piles is constructed based on graph nodes, node features, and edge connection relationships.

6. The method for coordinated early warning of deformation stress in soft soil around piles according to claim 1, characterized in that... The obtained early warning results for soft soil around the pile include: Embedding learning is performed on the deformation-stress co-location graph based on graph neural network to obtain a low-dimensional embedding representation of each graph node; The Euclidean distance between each graph node is calculated based on the low-dimensional embedding representation, and the outlier of each graph node is calculated using the Euclidean distance. Cluster analysis based on anomalies was used to delineate potential anomalous areas in the soft soil around the piles. The outlier value of the outlier region is calculated by the outlier degree of all graph nodes within the outlier region; By comparing the outliers with the preset graded early warning standards, the early warning results for soft soil around the piles are obtained.

7. A collaborative early warning system for deformation stress in soft soil around piles, characterized in that, include: The acquisition unit is used to acquire monitoring data of the soft soil around the pile based on the monitoring points. The monitoring data includes pile deformation data, pile stress data, soil deformation data, soil stress data and environmental data. The monitoring points include pile monitoring points and soil monitoring points. The feature extraction unit is used to extract features from the monitoring data. Based on the differences in information contribution and sensitivity selection, monitoring indicators reflecting the deformation and stress evolution of the soft soil around the pile are obtained. The building blocks are used to construct models based on monitoring data and indicators. By considering risk assessment and spatial heterogeneity, the deformation-stress synergistic spectrum of the soft soil around the pile is obtained, including: A mechanical response prediction model reflecting the deformation-stress coupling relationship of soft soil around piles was constructed based on monitoring data. The mechanical response prediction model is used to obtain the predicted mechanical response value of each soil monitoring point at a future time. The predicted mechanical response and monitoring indicators are input into the risk assessment model for risk assessment, and the assessment results of each soil monitoring point in the soft soil around the pile are obtained. The assessment results include risk score, status label and evolution trend. Using the assessment results as node characteristics and combining the spatial distribution relationship between soil monitoring points, a deformation-stress synergistic map of the soft soil around the pile is constructed. The early warning unit is used to identify anomalies in soft soil around piles based on deformation-stress synergistic maps. It obtains abnormal areas through embedding learning and cluster analysis, and performs hierarchical early warnings based on the abnormal areas to obtain early warning results for soft soil around piles.

8. A collaborative early warning device for deformation stress in soft soil around piles, characterized in that, include: Memory, used to store computer programs; A processor, configured to implement the steps of the collaborative early warning method for deformation stress of soft soil around piles as described in any one of claims 1 to 6 when executing the computer program.

9. A readable storage medium, characterized in that: The readable storage medium stores a computer program that, when executed by a processor, implements the steps of the pile perimeter soft soil deformation stress collaborative early warning method as described in any one of claims 1 to 6.

Images