Subgrade graded filling construction settlement prediction method and system

Through multi-source dynamic monitoring and hybrid prediction models, the problem of difficult to evaluate the coupling impact of multi-source factors in traditional roadbed settlement prediction methods is solved, and the accurate prediction of roadbed settlement and the intelligent construction management is realized, which improves the real-time and visualization of the construction process.

CN120373131AInactive Publication Date: 2025-07-25JIANGSU MODERN SHUNING ENGINEERING CONSTRUCTION CO LTD
View PDF 0 Cites 7 Cited by

Patent Information

Application Number
CN202510538282.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-04-27
Publication Date
2025-07-25
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Traditional roadbed settlement prediction methods lack systematic integration of multi-source factors, making it difficult to evaluate the coupling effect of multi-source factors on settlement, resulting in insufficient real-time and visualization of construction adjustments, unable to adapt to dynamic changes in different geological conditions and construction conditions, low data transmission and processing efficiency, and unable to meet the needs of dynamic monitoring and immediate response during construction.

Method used

Multi-source dynamic monitoring, working condition classification model, hybrid prediction model and decision support system are adopted, and through data perception module, edge computing module, central processor and digital twin platform, multi-source information fusion and real-time monitoring are realized, complex working conditions are identified, hybrid prediction models are established, decision-making plans are generated and visualized.

Benefits of technology

It improves the accuracy of roadbed settlement prediction and the intelligent level of construction management, realizes adaptability to complex working conditions, ensures safety and economy, reduces construction period delays, and provides visual support for the construction process.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120373131A_ABST
    Figure CN120373131A_ABST
Patent Text Reader

Abstract

The invention discloses a subgrade graded filling construction settlement prediction method and system, and relates to the technical field of subgrade construction.The subgrade graded filling construction settlement prediction system comprises a data sensing module, an edge calculation module, a central processor, an emergency decision module and a digital twinborn platform; non-linear working condition classification is carried out through an edge calculation module, complex working conditions are effectively identified, and the accuracy of working condition classification is improved; a mechanism that environmental factors indirectly influence settlement is disclosed through a central processor, the synergistic effect of environmental and geological factors is captured, then a hybrid prediction model is established to achieve quantitative evaluation of the roadbed filling state, and the adaptability to complex working conditions is improved; and then dynamic balance of settlement control and resource allocation is realized through an emergency decision module, visual support is provided for construction management through digital twinning and 3D modeling, decision efficiency and transparency are improved, and the method has important engineering application value and technical leading.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of subgrade construction, and particularly to a method and system for predicting settlement during hierarchical filling construction of subgrade. Background Art

[0002] In the field of subgrade construction, settlement control is a key link to ensure project quality and safety. It can guarantee the integrity of the subgrade structure, avoid catastrophic damage, ensure the long-term stability of project functions, avoid safety risks, prevent secondary disasters, improve project economy, and avoid waste of resources.

[0003] However, the data acquisition and processing process of traditional subgrade settlement prediction methods has the defects of isolation, difficulty in evaluating the coupling effects of multi-source factors on settlement, and multiple objectives of settlement control, construction period, cost and safety, resulting in insufficient real-time and visualization of subgrade construction adjustment.

[0004] Among them, because traditional methods only rely on single settlement monitoring data and lack systematic integration of geological factors (such as pore water pressure, underground anomalies) and environmental factors (such as rainfall, temperature, wind speed), traditional models are mostly based on linear assumptions and ignore the non-linear interactions of pore water pressure changes in soil, characteristics of underground anomalies and environmental factors, such as the complex effects of soil softening caused by rainfall and freeze-thaw cycles caused by temperature on settlement, making it difficult to comprehensively reveal the coupling effects of multi-source factors on settlement.

[0005] Moreover, lacking intelligent decision support based on real-time working conditions, construction adjustments (such as material ratio, filling rate) rely on manual experience, making it difficult to balance multiple objectives of settlement control, construction period, cost and safety, unable to adapt to dynamic changes in different geological conditions and construction working conditions, and the resulting defects of insufficient real-time and visualization, low data transmission and processing efficiency, unable to meet the needs of dynamic monitoring and immediate response during construction, prone to lag in risk warning, unable to intuitively present the settlement evolution trend and multi-factor influence path during subgrade construction, increasing the difficulty of project management.

[0006] In view of the above technical defects, a solution is proposed now. Summary of the Invention

[0007] The purpose of the present invention is to solve the defects that the data acquisition and processing process of traditional subgrade settlement prediction methods has isolation, difficulty in evaluating the coupling effects of multi-source factors on settlement, and multiple objectives of settlement control, construction period, cost and safety, resulting in insufficient real-time and visualization of subgrade construction adjustment.

[0008] To achieve the above purpose, the present invention adopts the following technical solutions:

[0009] A method for predicting the settlement of subgrade graded filling construction, comprising the following steps:

[0010] Step 1, multi-source dynamic monitoring and collection of multi-source information of the subgrade;

[0011] Step 2, extraction of working condition characteristics and establishment of a working condition classification model: by preprocessing the multi-source information of the subgrade, the real-time working condition state is determined;

[0012] Step 3, establishment of a hybrid prediction model and generation of the prediction result of subgrade settlement: by building an environmental physical correction sub-model, the environmental characteristic influence function is output; and by building a multi-source data-driven sub-model, the multi-source characteristic correlation function is output; and then the prediction result of subgrade settlement is generated by fusion;

[0013] Step 4, receiving the prediction result of subgrade settlement and generating a decision-making plan: establishing a state space and an action space, outputting the objective function of the dynamic optimization plan, so as to generate the corresponding decision-making plan and management instructions;

[0014] Step 5, carrying out 3D subgrade visualization modeling: by simulating the subgrade settlement trend under the decision-making plan, the subgrade construction state is simulated, and the visualization display of the subgrade graded filling construction settlement is realized.

[0015] Furthermore, the multi-source information of the subgrade includes settlement data, geological data and environmental data;

[0016] The settlement data includes the settlement amounts of N0 observation points; the geological data includes the pore water pressure of the soil mass and the parameters of underground anomalies; the environmental data includes rainfall, temperature and wind speed.

[0017] Furthermore, the specific process of building the working condition classification model is as follows:

[0018] By preprocessing the multi-source information of the subgrade, the real-time working condition state is determined;

[0019] Data cleaning is carried out on the collected multi-source information of the subgrade, and then feature selection is carried out on the cleaned multi-source information of the subgrade: the correlation between geological data, environmental data and settlement data is analyzed through the Pearson correlation coefficient;

[0020] The multi-source information of the subgrade is standardized, and settlement characteristics, geological characteristics and environmental characteristics are extracted;

[0021] The settlement characteristics include settlement difference, settlement rate and settlement acceleration; the geological characteristics include the difference in pore water pressure of the soil mass at different observation points at the same depth, and the distance between the underground anomaly and the observation point; the environmental characteristics include daily cumulative rainfall, rainfall intensity, daily temperature difference and daily maximum wind speed;

[0022] Construct the spatial distribution matrices of the settlement field and the geological field, and the time distribution matrix of the environmental field respectively through the settlement characteristics, geological characteristics, and environmental characteristics;

[0023] Concatenate the matrices of the settlement field, geological field, and environmental field into a unified feature vector. The dimension M of the unified feature vector is the sum of the number of settlement characteristics, geological characteristics, and environmental characteristics;

[0024] Mark the time dimension as T and the spatial dimension as S. Then mark the spatio-temporal integration matrix of the unified feature vector as a three-dimensional tensor R. The dimension of the three-dimensional tensor R is T×S×M;

[0025] Then build a support vector machine (SVM) model through the three-dimensional tensor R and output the corresponding working condition labels.

[0026] Furthermore, the specific process of building the support vector machine (SVM) model is as follows:

[0027] Input the historical data of the three-dimensional tensor R, and perform working condition classification and marking on the historical data to obtain working condition labels. The working condition labels include normal working condition C1, rainfall saturation C2, freeze-thaw softening C3, and load mutation C4;

[0028] Randomly divide the historical data into a training dataset, a validation dataset, and a test dataset;

[0029] Perform non-linear classification processing on the training dataset using the radial basis function (RBF) kernel to obtain the predicted working condition labels; use the Bayesian optimization algorithm to select the optimal value of the regularization parameter C for the validation dataset, and adjust the kernel parameter γ through cross-validation; calculate the classification accuracy, recall rate, and F1-score values through the test dataset to verify the model of the decision function, and substitute the real-time collected multi-source information of the subgrade into the decision function and output the corresponding working condition labels.

[0030] Furthermore, the specific process of establishing the hybrid prediction model is as follows:

[0031] Build an environmental physical correction sub-model to analyze the single-source influence degree of environmental factors on geological factors and output the environmental characteristic influence function;

[0032] Conduct regression analysis on environmental characteristics and geological characteristics, and build linear regression models between soil pore water pressure and rainfall, temperature, and wind speed respectively;

[0033] Furthermore, conduct multi-factor coupling influence analysis on the linear regression models between soil pore water pressure and rainfall, temperature, and wind speed respectively, so as to establish a multiple linear regression model, analyze the mechanism by which environmental factors indirectly affect subgrade settlement by changing soil physical properties, and output the environmental characteristic influence function F1;

[0034] By building a multi-source data-driven sub-model, analyze the multi-source influence degree of environmental factors and geological factors on subgrade settlement, introduce non-linear terms and interaction terms, analyze the synergy between environmental factors and geological factors on subgrade settlement, and output the multi-source feature correlation function F2;

[0035] Furthermore, fuse the environmental feature influence function F1 and the multi-source feature correlation function F2 to obtain the subgrade settlement prediction index X, set the evaluation interval of the subgrade settlement prediction index X, and generate the prediction result of the subgrade settlement through interval comparison.

[0036] Furthermore, the specific generation process of the decision-making plan is as follows:

[0037] Establish the state space and action space: Among them, the state space includes the working condition type, settlement deviation, and resource constraints; the action space includes adjusting the material ratio, adjusting the filling rate, pausing construction, strengthening the foundation, and constructing in sections and layers to disperse the load;

[0038] Thereby output the objective function H of the dynamic optimization plan, then generate the corresponding decision-making plan and management instructions, mark the optimization plan corresponding to the optimal objective function value Hmin as the optimal decision-making plan, and generate and output the corresponding management instructions according to the action space vector corresponding to the optimal decision-making plan.

[0039] A subgrade graded filling construction settlement prediction system includes a data perception module, an edge computing module, a central processor, an emergency decision-making module, and a digital twin platform. Among them, the data perception module, the edge computing module, the central processor, the emergency decision-making module, and the digital twin platform are communicatively connected, and this system applies the above-mentioned subgrade graded filling construction settlement prediction method;

[0040] The data perception module is used for multi-source dynamic monitoring and collecting multi-source information of the subgrade;

[0041] The edge computing module is used for extracting working condition features, building a working condition classification model, and preprocessing the multi-source information of the subgrade to determine the real-time working condition state;

[0042] The central processor is used for building an environmental physical correction sub-model and a multi-source data-driven sub-model, respectively outputting the environmental feature influence function and the multi-source feature correlation function, and then establishing a hybrid prediction model, fusing the environmental feature influence function and the multi-source feature correlation function, and generating the prediction result of the subgrade settlement;

[0043] The emergency decision-making module is used for receiving the prediction result of the subgrade settlement, establishing the state space and the action space, outputting the objective function of the dynamic optimization plan, and thus generating the decision-making plan and management instructions;

[0044] The digital twin platform is used for 3D subgrade visualization modeling, simulating the subgrade settlement trend under the decision-making scheme, thereby simulating the subgrade construction state, and realizing the visual display of the subgrade graded filling construction settlement.

[0045] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are as follows:

[0046] The present invention improves the multi-source data fusion and real-time monitoring capabilities through the data perception module, and classifies non-linear working conditions through the edge computing module, effectively identifying complex working conditions such as normal working conditions, rainfall saturation, freeze-thaw softening, and load mutation, improving the accuracy of working condition classification;

[0047] The present invention builds an environmental physical correction sub-model and a multi-source data-driven sub-model through the central processor, reveals the mechanism by which environmental factors indirectly affect settlement by changing the physical properties of the soil, and captures the synergistic effect of environmental and geological factors, and then establishes a hybrid prediction model to achieve a quantitative assessment of the subgrade filling state, improving the adaptability to complex working conditions;

[0048] The present invention establishes a state space and an action space through multi-objective optimization decision-making, and sets the objective function to be minimized to achieve the dynamic balance of settlement control and resource allocation, reducing the construction period delay while ensuring safety, and then providing visual support for construction management through digital twin and 3D modeling, improving decision-making efficiency and transparency.

[0049] In summary, the present invention constructs a full-process closed-loop of "monitoring - analysis - prediction - decision - visualization" through technological innovation, significantly improving the accuracy of subgrade settlement prediction and the intelligent level of construction management, and having important engineering application value and technological leadership. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 The step schematic diagram showing the working process of the present invention;

[0051] Figure 2 The connection schematic diagram showing the system modules of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0052] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts belong to the scope of protection of the present invention.

[0053] Embodiment 1:

[0054] As Figure 1 - Figure 2As shown in the figure, a subgrade graded filling construction settlement prediction system includes a data perception module, an edge computing module, a central processor, an emergency decision-making module, and a digital twin platform. Among them, the data perception module, the edge computing module, the central processor, the emergency decision-making module, and the digital twin platform are communicatively connected;

[0055] The working steps are as follows:

[0056] S1. The data perception module is used for multi-source dynamic monitoring to collect multi-source information of the subgrade. The specific process is as follows:

[0057] The multi-source information of the subgrade includes settlement data, geological data, and environmental data;

[0058] S1-1. The settlement data includes the settlement amounts of N0 observation points;

[0059] By setting control points at stable positions around the subgrade and arranging N0 observation points on the subgrade, using a total station to measure the horizontal angle, vertical angle, and inclined distance between the control points and the observation points, calculating the three-dimensional coordinates of the observation points, and determining the settlement amount by comparing the coordinate data of the observation points in different periods;

[0060] Among them, the total station, that is, the total station electronic tachymeter, is a high-tech measuring instrument integrating optics, mechanics, and electronics, and is a surveying and mapping instrument system integrating the functions of horizontal angle, vertical angle, distance, and height difference measurement;

[0061] S1-2. The geological data includes soil pore water pressure and underground anomaly parameters;

[0062] Monitoring the soil pore water pressure at different depths of the observation points through a pore water pressure gauge; Scanning underground anomalies such as cavities and soft soil interlayers through a geological radar and recording the underground anomaly parameters, including the size and location of the underground anomalies;

[0063] Monitoring the geological radar reflection signal, removing noise interference through filtering, and then identifying effective anomalies based on the threshold method: presetting the interval of the geological radar reflection signal intensity of normal soil, and when the monitored geological radar reflection signal intensity is not within the preset interval, it is determined that there is an underground anomaly here;

[0064] Then, by setting the interval of the radar reflection signal intensity of any underground anomaly a, when the monitored geological radar reflection signal intensity is within the interval of the radar reflection signal intensity of the corresponding underground anomaly a, it is determined that the underground anomaly a exists;

[0065] Through the processing and analysis of existing professional software, obtaining the reflection wave image characteristics, including the waveform, amplitude, and arrival time of the reflection wave, so as to determine the position and size of the anomaly. The collection process of geological data does not fall within the protection scope of the present invention and will not be elaborated here;

[0066] S1-3, environmental data include rainfall, temperature and wind speed;

[0067] Periodic data on rainfall, temperature, and wind speed are collected through weather stations;

[0068] The multi-source information of the roadbed is synchronized through LoRa network transmission, and the timestamps are aligned and pre-processed.

[0069] S2, the edge computing module extracts working condition characteristics, builds a working condition classification model, and pre-processes the multi-source information of the roadbed to determine the real-time working condition status. The specific process is as follows:

[0070] S2-1, data cleaning of the collected roadbed multi-source information: linear interpolation method is used to fill the missing values of the roadbed multi-source information, and data points exceeding the mean ±3 times the standard deviation are identified based on the 3σ principle, marked as abnormal samples and eliminated;

[0071] S2-2, feature selection of multi-source information of the cleaned roadbed: analysis of the correlation between geological data, environmental data and settlement data through the Pearson correlation coefficient;

[0072] S2-201, settlement data includes settlement amounts at N0 observation points;

[0073] Standardize the settlement data and extract settlement characteristics through the settlement data. The settlement characteristics include settlement difference, settlement rate and settlement acceleration.

[0074] The settlement difference refers to the difference in settlement between adjacent observation points; the settlement rate refers to the settlement per unit time; the settlement acceleration refers to the rate of change of the settlement rate;

[0075] The functions of the settlement data feature are:

[0076] Settlement difference: reflects the settlement difference of different parts of the roadbed. Too large settlement difference may cause road cracking and roadbed instability. It is a key indicator for evaluating uneven settlement of roadbed.

[0077] Settlement rate: reflects the speed of roadbed settlement. A sudden increase in settlement rate may indicate potential problems in the roadbed, such as soil sliding, foundation failure, etc.

[0078] Settlement acceleration: reflects the changing trend of settlement rate and can provide early warning of the accelerated development of roadbed settlement.

[0079] S2-202, geological data including soil pore water pressure and underground anomaly body parameters;

[0080] Standardize the geological data and extract geological features from the geological data. The geological features include the difference in soil pore water pressure at different observation points at the same depth and the distance between the underground anomaly and the observation point.

[0081] S2-203, the environmental data includes rainfall, temperature and wind speed.

[0082] Standardize the environmental data and extract environmental features from the environmental data. The environmental features include daily cumulative rainfall, rainfall intensity, daily temperature difference and daily maximum wind speed.

[0083] Among them, the Z-score standardization method is adopted for standardization to dimensionless the data of various categories.

[0084] S2-3, construct the spatial distribution matrix of the settlement field and the geological field, and the time distribution matrix of the environmental field respectively through the settlement features, geological features and environmental features.

[0085] Stitch the matrices of the settlement field, geological field and environmental field into a unified feature vector. The dimension M of the unified feature vector is the sum of the number of settlement features, geological features and environmental features.

[0086] Mark the time dimension as T and the spatial dimension as S, then mark the spatio-temporal integration matrix of the unified feature vector as the three-dimensional tensor R, and the dimension of the three-dimensional tensor R is T×S×M.

[0087] S2-4, build a support vector machine SVM model. The specific process is as follows:

[0088] Input the historical data of the three-dimensional tensor R, and perform working condition classification and marking on the historical data to obtain the working condition labels. The working condition labels include normal working condition C1, rainfall saturation C2, freeze-thaw softening C3 and load mutation C4. Obtain the three-dimensional tensor R corresponding to any working condition label to establish a template sequence.

[0089] Randomly divide the historical data into training dataset, validation dataset and test dataset.

[0090] Perform non-linear classification processing on the training dataset using the radial basis function RBF kernel, and construct a decision function by maximizing the classification margin and minimizing the classification error to obtain the predicted working condition label.

[0091] Use the Bayesian optimization algorithm to select the optimal value of the regularization parameter C for the validation dataset, and adjust the kernel parameter γ through cross-validation to avoid underfitting and overfitting and improve the generalization performance of the model.

[0092] Calculate the classification accuracy, recall rate and F1-score value through the test dataset to verify the decision function model, and substitute the real-time collected multi-source information of the subgrade into the decision function and output the corresponding working condition label.

[0093] S3. Build an environmental physical correction sub-model and a multi-source data-driven sub-model for the central processing unit, respectively output an environmental feature influence function and a multi-source feature correlation function, and then establish a hybrid prediction model. Integrate the environmental feature influence function and the multi-source feature correlation function to generate a prediction result of subgrade settlement;

[0094] The specific process of establishing the hybrid prediction model is as follows:

[0095] S3-1. By building an environmental physical correction sub-model, analyze the single-source influence degree of environmental factors on geological factors, and output an environmental feature influence function;

[0096] Through regression analysis of environmental features and geological features, obtain the Pearson correlation coefficient, extract the main influencing variables with high correlation, including soil pore water pressure, rainfall, temperature, and wind speed, and build linear regression models between soil pore water pressure and rainfall, temperature, and wind speed respectively;

[0097] Furthermore, through the principal component analysis method and weighted fusion, conduct a multi-factor coupling influence analysis on the linear regression models between soil pore water pressure and rainfall, temperature, and wind speed respectively, so as to establish a multiple linear regression model and output an environmental feature influence function;

[0098] Mark the soil pore water pressure as Ph, and mark the rainfall, temperature, and wind speed as R(τ), T(t), and V(τ) respectively;

[0099] Then mark the environmental feature influence function as F1:

[0100]

[0101] Among them, e1, e2, and e3 are the weight coefficients of rainfall, temperature, and wind speed respectively. The weight coefficients are preset through the principal component analysis method and combined with a large amount of data training.

[0102] S3-2. By building a multi-source data-driven sub-model, analyze the multi-source influence degree of environmental factors and geological factors on subgrade settlement, and output a multi-source feature correlation function:

[0103] Mark the environmental feature vector as Environmental features include daily cumulative rainfall E1, rainfall intensity E2, daily temperature difference E3, and daily maximum wind speed E4;

[0104] Mark the geological feature vector as Geological features include the difference in soil pore water pressure G1 and the distance G2 from the underground anomaly to the observation point;

[0105] Label the settlement feature vector as The settlement features include settlement difference D1, settlement rate D2, and settlement acceleration D3;

[0106] Then label the multi-source feature correlation function as F2:

[0107]

[0108] Wherein, is the linear influence coefficient vector; is the non-linear term coefficient vector, is the interaction term coefficient vector, is the error term vector; is the environmental feature vector is the element square operation of; is the geological feature vector is the element square operation of; is the environmental feature vector and the geological feature vector is the element cross multiplication of; The influence of the factor's own change rate is evaluated through the non-linear term, and the synergistic effect of environmental and geological factors is reflected through the interaction term;

[0109] By using the least squares method for the linear term minimize the mean square error MSE between the predicted value and the actual value of the settlement feature, so as to obtain the linear influence coefficient vector Iteratively optimize the quadratic term and the interaction term through the gradient descent method to further reduce the mean square error MSE, so as to obtain the non-linear term coefficient vector and the interaction term coefficient vector

[0110] S3-3, and then fuse through the environmental feature influence function F1 and the multi-source feature correlation function F2 to obtain the subgrade settlement prediction index X: Wherein, refers to the multiplication of the results of the environmental feature influence function F1 and the multi-source feature correlation function F2. The higher the influence degrees of the environmental feature influence function F1 and the multi-source feature correlation function F2, the higher the subgrade settlement prediction index;

[0111] Set the evaluation interval of the subgrade settlement prediction index X, and generate the prediction result of the subgrade settlement through interval comparison: The prediction results are divided into stable, critical, and unstable subgrade filling states; The higher the subgrade settlement prediction index, the more unstable the prediction result of the subgrade settlement is evaluated; On the contrary, the more stable the prediction result of the subgrade settlement is evaluated.

[0112] S3-4, it should be noted that:

[0113] S3-401, Analysis of the Influence of Environmental Data on Geological Data, is specifically described as follows:

[0114] Influence of rainfall on geological data: Changes in rainfall will directly affect the water content in the soil mass, and thus have a significant impact on the pore water pressure in the soil mass. When rainfall increases, water seeps into the soil mass through the ground surface, the amount of water in the soil pores increases, and the pore water pressure will correspondingly rise; at the same time, for underground anomalies, changes in the water content of the surrounding soil mass may cause changes in its physical properties. For example, the strength of a soft soil interlayer decreases after absorbing water.

[0115] Influence of temperature on geological data: Changes in temperature will cause the soil mass to expand and contract thermally, resulting in changes in the pore structure of the soil mass. When the temperature rises, the soil particles expand, which may reduce the pore volume and increase the pore water pressure; conversely, when the temperature drops, the soil particles contract, the pore volume increases, and the pore water pressure may decrease.

[0116] Influence of wind speed on geological data: Wind speed indirectly affects the water content in the soil mass by influencing the evaporation on the ground surface. A higher wind speed will accelerate the evaporation of surface water, resulting in a decrease in the water content of the shallow soil layer and a decrease in the pore water pressure, which may gradually transfer to the deep soil layer over time.

[0117] S3-402, Analysis of the Influence of Multi-source Factors on Subgrade Settlement, is specifically described as follows:

[0118] Influence of environmental factors:

[0119] Rainfall: A large amount of rainfall will increase the water content of the subgrade soil mass, causing the soil mass to soften and the shear strength to decrease, thereby triggering subgrade settlement; the infiltration of rainwater may also cause the groundwater level to rise, weakening the bearing capacity of the foundation and aggravating the settlement; at the same time, rainwater scouring may cause the instability of the subgrade slope, indirectly affecting the overall settlement state of the subgrade.

[0120] Temperature: Temperature changes cause the soil mass to expand and contract thermally. Repeated temperature fluctuations will damage the soil structure, resulting in uneven subgrade settlement; in a low-temperature environment, the water in the soil mass freezes and expands, and the soil density decreases after thawing, which will also cause settlement.

[0121] Wind speed: Wind speed affects the water content of the soil mass by influencing the evaporation amount, and thus affects the subgrade settlement; strong winds accelerate the evaporation of water, causing the surface soil layer of the subgrade to dry shrink, generating cracks and settlement.

[0122] Influence of geological factors:

[0123] Soil pore water pressure: An increase in pore water pressure will reduce the effective stress of the soil mass, causing the soil mass to undergo compressive deformation and resulting in subgrade settlement; changes in pore water pressure at different depths may also cause uneven settlement.

[0124] Underground anomalies: Underground anomalies such as cavities and soft soil interlayers will weaken the bearing capacity of the foundation, resulting in large settlements of the roadbed under the action of loads; different positions and sizes of the anomalies will have different degrees of influence on the settlement of the roadbed.

[0125] S4. The emergency decision-making module receives the prediction results of the roadbed settlement and generates a decision-making plan: establish a state space and an action space, output the objective function of the dynamic optimization plan, so as to generate a decision-making plan and management instructions;

[0126] S4-1. Establish a state space and an action space:

[0127] The state space includes the working condition type, settlement deviation, and resource constraints;

[0128] Among them, the working condition type refers to the working condition label; the settlement deviation refers to the predicted settlement amount of the roadbed; the resource constraints include cost, construction period, and safety accidents;

[0129] The action space includes adjusting the material ratio, adjusting the filling rate, suspending construction, strengthening the foundation, and constructing in sections and layers to disperse the load;

[0130] S4-2. Output the objective function H of the dynamic optimization plan:

[0131] Among them, Bj refers to the action space vector, including various preset optimization plans, ωi refers to the weight coefficient of the state space vector Zi, and the settlement deviation and resource constraints are integrated into the state space vector Zi, that is, the state space vector Zi includes settlement amount, cost, construction period, and safety accidents; the safety accident level is quantitatively rated and determined by industry technicians; by dynamically adjusting the action space vector Bj of the optimization plan, the objective function H is minimized and marked as the optimal objective function value Hmin;

[0132] S4-3. Then generate the corresponding decision-making plan and management instructions;

[0133] Mark the optimization plan corresponding to the optimal objective function value Hmin as the optimal decision-making plan, and generate and output the corresponding management instructions according to the action space vector corresponding to the optimal decision-making plan to achieve the balance among roadbed settlement, construction period, cost, and safety.

[0134] S5. The digital twin platform conducts 3D visualization modeling of the roadbed: By simulating the roadbed settlement trend under the decision-making plan, the construction state of the roadbed is simulated to realize the visual display of the settlement during the staged filling construction of the roadbed; among them, the digital twin platform conducts the whole-process monitoring from the collection of multi-source information of the roadbed to the data analysis and processing.

[0135] The working effects of the present invention are as follows:

[0136] The present invention improves the multi-source data fusion and real-time monitoring capabilities through a data perception module, and classifies non-linear working conditions through an edge computing module, effectively identifying complex working conditions such as normal working conditions, rainfall saturation, freeze-thaw softening, and load mutation, improving the accuracy of working condition classification;

[0137] The present invention builds an environmental physical correction sub-model and a multi-source data-driven sub-model through a central processor, reveals the mechanism by which environmental factors indirectly affect settlement by changing the physical properties of soil, and captures the synergistic effect of environmental and geological factors, and then establishes a hybrid prediction model to achieve a quantitative assessment of the subgrade filling state, improving the adaptability to complex working conditions;

[0138] The present invention establishes a state space and an action space through multi-objective optimization decision-making, and sets the objective function to be minimized to achieve the dynamic balance of settlement control and resource allocation, reducing the construction period delay while ensuring safety, and then providing visual support for construction management through digital twin and 3D modeling, improving decision-making efficiency and transparency.

[0139] In terms of engineering application value, the present invention significantly improves safety, optimizes economy, and enhances suitability. In summary, the present invention constructs a full-process closed loop of "monitoring - analysis - prediction - decision-making - visualization" through technological innovation, significantly improving the accuracy of subgrade settlement prediction and the intelligent level of construction management, and having important engineering application value and technological leadership.

[0140] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution.

[0141] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0142] The above is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution of the present invention and its inventive concept, makes equivalent substitutions or changes, and should be covered by the protection scope of the present invention.

Claims

1. A method for predicting construction settlement of subgrade graded filling, characterized in that: It includes the following steps: Step 1, perform multi-source dynamic monitoring and collect multi-source information of the subgrade; Step 2, extract working condition characteristics and build a working condition classification model: preprocess the multi-source information of the subgrade to determine the real-time working condition status; Step 3, establish a hybrid prediction model and generate the prediction result of subgrade settlement: build an environmental physical correction sub-model to output the environmental characteristic influence function; and build a multi-source data-driven sub-model to output the multi-source characteristic correlation function; then fuse to generate the prediction result of subgrade settlement; Step 4, receive the prediction result of subgrade settlement and generate a decision-making plan: establish a state space and an action space, output the objective function of the dynamic optimization plan, and thus generate the corresponding decision-making plan and management instructions; Step 5, perform 3D visualization modeling of the subgrade: simulate the subgrade settlement trend under the decision-making plan to simulate the subgrade construction status and realize the visual display of the settlement during the graded filling construction of the subgrade.

2. The method for predicting the construction settlement of subgrade graded filling according to claim 1, wherein: The multi-source information of the subgrade includes settlement data, geological data, and environmental data; The settlement data includes the settlement amounts of N0 observation points; the geological data includes the pore water pressure of the soil and the parameters of underground anomalies; the environmental data includes rainfall, temperature, and wind speed.

3. A method for predicting the settlement of subgrade construction by graded filling according to claim 2, characterized in that: The specific process of building the working condition classification model is as follows: Preprocess the multi-source information of the subgrade to determine the real-time working condition status; Perform data cleaning on the collected multi-source information of the subgrade, and then perform feature selection on the cleaned multi-source information of the subgrade: analyze the correlation between geological data, environmental data, and settlement data through the Pearson correlation coefficient; Standardize the multi-source information of the subgrade and extract settlement characteristics, geological characteristics, and environmental characteristics; The settlement characteristics include settlement difference, settlement rate, and settlement acceleration; the geological characteristics include the difference in pore water pressure of the soil at different observation points at the same depth and the distance between the underground anomaly and the observation point; the environmental characteristics include daily cumulative rainfall, rainfall intensity, daily temperature difference, and daily maximum wind speed; Construct the spatial distribution matrices of the settlement field and the geological field, and the time distribution matrix of the environmental field through the settlement characteristics, geological characteristics, and environmental characteristics respectively; Concatenate the matrices of the settlement field, geological field, and environmental field into a unified feature vector, and the dimension M of the unified feature vector is the sum of the settlement feature number, geological feature number, and environmental feature number; Mark the time dimension as T and the spatial dimension as S, then mark the spatio-temporal integration matrix of the unified feature vector as the three-dimensional tensor R, and the dimension of the three-dimensional tensor R is T×S×M; Then build a support vector machine (SVM) model through the three-dimensional tensor R and output the corresponding working condition labels.

4. A method for predicting the settlement of subgrade construction by graded filling according to claim 3, characterized in that: The specific process of building the support vector machine (SVM) model is as follows: Input the historical data of the three-dimensional tensor R and perform working condition classification marking on the historical data to obtain working condition labels, and the working condition labels include normal working condition C1, rainfall saturation C2, freeze-thaw softening C3, and load mutation C4; Randomly divide the historical data into a training dataset, a validation dataset, and a test dataset; The training dataset is processed by non - linear classification using the radial basis function (RBF) kernel to obtain the predicted working condition labels. The validation dataset is used to select the optimal value of the regularization parameter C by the Bayesian optimization algorithm, and the kernel parameter γ is adjusted through cross - validation. The classification accuracy, recall rate, and F1 - score values are calculated through the test dataset to verify the decision function model. Then, the multi - source information of the roadbed collected in real - time is substituted into the decision function to output the corresponding working condition labels.

5. A method for predicting the settlement during the construction of graded filling of subgrade according to claim 4, characterized in that: The specific process of establishing the hybrid prediction model is as follows: By building an environmental physical correction sub - model, analyze the single - source influence degree of environmental factors on geological factors, and output the environmental characteristic influence function. Through regression analysis of environmental characteristics and geological characteristics, and build linear regression models between soil pore water pressure and rainfall, temperature, and wind speed respectively. Furthermore, through multi - factor coupling influence analysis of the linear regression models between soil pore water pressure and rainfall, temperature, and wind speed respectively, a multiple linear regression model is established to analyze the mechanism by which environmental factors indirectly affect roadbed settlement by changing soil physical properties, and the environmental characteristic influence function F1 is output. By building a multi - source data - driven sub - model, analyze the multi - source influence degree of environmental factors and geological factors on roadbed settlement, introduce non - linear terms and interaction terms, and analyze the synergistic effect of environmental factors and geological factors on roadbed settlement, and output the multi - source characteristic correlation function F2. Furthermore, through the fusion of the environmental characteristic influence function F1 and the multi - source characteristic correlation function F2, the roadbed settlement prediction index X is obtained, and the evaluation interval of the roadbed settlement prediction index X is set, and the prediction result of roadbed settlement is generated through interval comparison.

6. A method for predicting the settlement during the construction of graded filling of subgrade according to claim 5, characterized in that: The specific generation process of the decision - making scheme is as follows: Establish the state space and action space: Among them, the state space includes working condition types, settlement deviation, and resource constraints; the action space includes adjusting material ratio, adjusting filling rate, pausing construction, strengthening the foundation, and constructing in sections and layers to disperse the load. Thus, the objective function H of the dynamic optimization scheme is output, and then the corresponding decision - making scheme and management instructions are generated. Mark the optimization scheme corresponding to the optimal objective function value Hmin as the optimal decision - making scheme, and generate and output the corresponding management instructions according to the action space vector corresponding to the optimal decision - making scheme.

7. A subgrade graded filling construction settlement prediction system, characterized in that: It includes a data perception module, an edge computing module, a central processor, an emergency decision - making module, and a digital twin platform. The data perception module, the edge computing module, the central processor, the emergency decision - making module, and the digital twin platform are communicatively connected. This system applies a method for predicting the settlement of graded roadbed filling construction according to any one of claims 1 - 6 above. The data perception module is used for multi - source dynamic monitoring to collect multi - source information of the roadbed. The edge computing module is used to extract working condition characteristics, build a working condition classification model, pre - process the multi - source information of the roadbed, and thus determine the real - time working condition state. The central processor is used to build an environmental physical correction sub - model and a multi - source data - driven sub - model, output the environmental characteristic influence function and the multi - source characteristic correlation function respectively, and then establish a hybrid prediction model, fuse the environmental characteristic influence function and the multi - source characteristic correlation function, and generate the prediction result of roadbed settlement. The emergency decision-making module is used to receive the prediction results of subgrade settlement, establish the state space and action space, and output the objective function of the dynamic optimization scheme, so as to generate the decision-making scheme and management instructions; The digital twin platform is used for 3D visualization modeling of the subgrade, simulating the subgrade settlement trend under the decision-making scheme, so as to simulate the subgrade construction state and realize the visual display of the settlement during the graded filling construction of the subgrade.

Citation Information

Cited By

  • Self-adaptive multi-source fusion roadbed intelligent monitoring and dynamic regulation and control system and method

    CN120804669A

  • An adaptive multi-source fusion roadbed intelligent monitoring and dynamic regulation system and method

    CN120804669B

  • Geological settlement monitoring method and system integrating deep learning and multi-source data

    CN120929778A

  • Geological subsidence monitoring method and system fusing deep learning and multi-source data

    CN120929778B

  • Building overall settlement calculation method and system based on multi-source data fusion

    CN121351245A