CN-120542632-B - Intelligent prediction method and system for differential settlement of expressway widened roadbed

CN120542632BCN 120542632 BCN120542632 BCN 120542632BCN-120542632-B

Abstract

The invention provides an intelligent prediction method and system for differential settlement of expressway widening roadbed, and relates to the technical field of expressway widening. And collecting data of the joint of the new road surface and the old road surface in real time, constructing a three-dimensional stress field model, identifying a weak area, generating a topological structure, combining pore water pressure gradient, mapping coordinates to filler particle diameter parameters, dynamically distributing creep parameter weights, and generating a creep cloud image. Extracting a strain rate characteristic value, establishing a traffic load-shear strength degradation correlation matrix, correcting a shear strength factor, constructing a settlement differential model, and predicting settlement distribution. And comparing and optimizing parameters by using triple residual errors, outputting a sedimentation early warning map when the similarity coefficient reaches the standard, and dividing the early warning grade. The invention fuses multisource data, improves early warning overlap ratio, and improves settlement prediction precision and prevention and control timeliness.

Inventors

SUN FENG
ZHAO CHUANSONG
WU XIAOJIE
WAN HANGYANG

Assignees

江苏省交通工程集团有限公司

Dates

Publication Date: 20260505
Application Date: 20250514

Claims (8)

1. An intelligent prediction method for differential settlement of a highway widening roadbed is characterized by comprising the following steps: Based on the sensor, collecting compaction degree distribution data, ground water level change data and traffic load time course curve of the joint of the new road surface and the old road surface in real time, constructing a three-dimensional stress field model with a weak area marking layer, and synchronously outputting stress state parameters of grid nodes, wherein the grid nodes are basic units of the three-dimensional stress field model; Mapping the weak area marking layer into the filler particle size distribution parameters, dynamically distributing creep parameter weights based on the weak area marking layer coordinates, and generating a creep cloud chart by combining stress state parameters, wherein the creep cloud chart comprises time-varying strain rate distributing strips overlapped with the weak area marking layer; the creep cloud image is matched with grid nodes of a three-dimensional stress field model, strain rate characteristic values corresponding to a weak area marking layer are extracted, a traffic load-shear strength degradation correlation matrix taking a ground water level fluctuation amplitude as a constraint condition is established, and the correlation matrix is dynamically corrected by adopting a double-parameter reduction algorithm based on weak area water content sensitivity; Performing coupling operation on the corrected incidence matrix and stress state parameters, constructing a settlement differential model containing a weak area viscoelastic-plastic constitutive equation set, and generating a settlement distribution map fused with a weak area marking layer space topological structure through space-time convolution; The method comprises the steps of synchronously optimizing the spatial resolution and creep parameter weight of a weak area marking layer through triple residual error comparison of a three-dimensional stress field model, a creep cloud image and a settlement distribution diagram, wherein the triple residual error comprises a stress residual error of the three-dimensional stress field model, a strain residual error of the creep cloud image and a displacement residual error of the settlement distribution diagram, the residual error is obtained based on comparison of actual measurement data acquired by a sensor in real time and a model output value, and when the spatial overlap ratio of the weak area marking layer and an actual measurement settlement abnormal area reaches a set threshold value, a pavement settlement early warning map is output.
2. The intelligent prediction method for the differential settlement of the expressway widened roadbed is characterized by comprising the steps of constructing a three-dimensional stress field model, carrying out multi-source heterogeneous data fusion on sensor data, establishing a three-dimensional stress field model which is divided by a finite element grid and takes a weak area marking layer as a center, wherein the weak area marking layer is generated by adopting the geometric center which is lower than a set threshold area in compactness distribution data as a seed point, carrying out spatial clustering analysis by combining second derivative extreme points of stress state parameters, and generating a weak area three-dimensional topological structure with thickness gradient characteristics, wherein the three-dimensional stress field model is provided with a grid size which is one fifth to one third of that of a conventional area at the boundary of the weak area marking layer, and carrying out dynamic loading on pore water pressure gradient caused by underground water level change by adopting cubic spline interpolation, so as to form a three-dimensional tensor field containing the stress state parameters.
3. The intelligent prediction method for the differential settlement of the expressway widened roadbed is characterized in that when the creep cloud image is constructed by mapping the space coordinates of a weak area marking layer to filler particle size distribution parameters, the nonlinear mapping relation between the coarse particle content in a filler grading curve and the creep parameters is established by adopting coordinate transformation based on interpolation, wherein the dynamic distribution of the creep parameter weight is realized by defining Euclidean distance functions from nodes in the weak area marking layer to a nearest coarse particle enrichment area, and the specific expression is as follows: Wherein w i is creep parameter weight of a node i, lambda is grading sensitivity coefficient, x i is space coordinate of the node i in a weak area marking layer, x c is mass center coordinate of a coarse particle enrichment area, d max is maximum radial dimension of the weak area, the time-varying strain rate distribution belt is generated, a bias stress tensor component in stress state parameters is coupled with the weight coefficient and then input into a viscoelastic constitutive equation, and anisotropic creep strain increment is calculated through an implicit time integration algorithm.
4. The intelligent prediction method for the highway widening subgrade differential settlement according to claim 1, wherein the construction of the correlation matrix is characterized in that a two-dimensional matrix taking a ground water level fluctuation amplitude as a row vector and traffic load spectral density as a column vector is established by extracting a main strain rate gradient and a shear strain rate direction angle in a strain rate characteristic value, and the calculation of a shear strength degradation factor eta adopts a double-parameter reduction algorithm: Wherein eta 0 is initial shear strength, beta 1 is moisture content sensitivity coefficient, beta 2 is load action frequency sensitivity coefficient, alpha and gamma are fitting parameters, delta h is ground water level fluctuation amplitude, S (f) is traffic load spectrum density, the dynamic correction process is to adjust the value interval of beta 1 in real time by constructing the negative index relation between the moisture content of a weak area and the suction force of a matrix, and when the moisture content exceeds the plastic limit, a double-parameter reduction algorithm is activated.
5. The intelligent prediction method for the differential settlement of the expressway widened roadbed according to claim 1 is characterized in that the settlement differential model is constructed by taking a degradation factor in an incidence matrix as an input parameter to establish a viscoelastic-plastic constitutive equation system considering strain hardening and softening: ; wherein x is the back stress, Y is the yield stress, For relaxation time, C and n are material constants, σ is the stress tensor, t is time, E is the elastic modulus, In order to achieve a total strain rate, And the space-time convolution adopts a deep neural network with a cavity convolution kernel, the space topological structure of the weak area marking layer is encoded into an adjacent matrix, long-range space correlation characteristics are extracted through the graph convolution layer, and finally a sedimentation distribution diagram fused with traffic load time-course phase information is output.
6. The intelligent prediction method for highway widening subgrade differential settlement according to claim 1, which is characterized in that the triple residual comparison is carried out by defining a stress residual R σ of a three-dimensional stress field model, a strain residual R ϵ of a creep cloud picture and a displacement residual R u of a settlement distribution diagram, and calculating a comprehensive residual by adopting a weighted Euclidean distance formula: And synchronously updating the spatial resolution parameter and creep parameter weight of the weak area marking layer by adopting a conjugate gradient method in the optimization process, and triggering a spatial coincidence degree threshold condition when the similarity coefficient of the weak area marking layer and the actually measured abnormal area exceeds 0.75.
7. The intelligent prediction method for the differential settlement of the expressway widened roadbed is characterized in that the construction of the settlement pre-warning map is characterized in that a prediction result of the settlement pre-warning map is aligned with historical monitoring data in a space-time mode, HSV color space coding is adopted to generate a multi-layer superposition map with pre-warning grades, and pre-warning grade division is determined according to the ratio of the maximum settlement rate in a weak area marking layer to an allowable value.
8. A system for implementing an intelligent prediction method of differential settlement of a highway widening subgrade as set forth in any one of claims 1-7, comprising: The multi-source data acquisition unit is used for acquiring compaction degree distribution data, ground water level change data and traffic load time course curve of the joint of the new road surface and the old road surface in real time based on the sensor, constructing a three-dimensional stress field model with a weak area marking layer, and synchronously outputting stress state parameters of each grid node by the three-dimensional stress field model; The creep analysis unit is used for mapping the weak area marking layer into the filler particle size distribution parameters, dynamically distributing creep parameter weights based on the weak area marking layer coordinates, and generating a creep cloud chart by combining stress state parameters, wherein the creep cloud chart comprises time-varying strain rate distribution strips overlapped with the weak area marking layer; The coupling correction unit is used for extracting a strain rate characteristic value corresponding to a weak area marking layer through matching the creep cloud chart with grid nodes of the three-dimensional stress field model, and establishing a traffic load-shear strength degradation correlation matrix taking the fluctuation amplitude of the underground water level as a constraint condition, wherein the correlation matrix is dynamically corrected by adopting a double-parameter reduction algorithm based on the sensitivity of the water content of the weak area; The settlement prediction unit is used for carrying out coupling operation on the corrected incidence matrix and the stress state parameter, constructing a settlement differential model containing a weak area viscoelastic-plastic constitutive equation set and generating a settlement distribution map fused with a weak area marking layer space topological structure through space-time convolution; the system synchronously optimizes the spatial resolution and creep parameter weight of the weak area marking layer through triple residual error comparison of the three-dimensional stress field model-creep cloud chart-settlement distribution chart, and outputs a pavement settlement early warning map when the spatial overlap ratio of the weak area marking layer and the actually measured settlement abnormal area reaches a set threshold value.

Description

Intelligent prediction method and system for differential settlement of expressway widened roadbed Technical Field The invention relates to the technical field of expressway widening, in particular to an intelligent prediction method and system for expressway widening roadbed differential settlement. Background In expressway widening engineering, prediction and control of differential settlement of roadbed are dependent on traditional monitoring means and experience models for a long time. In the prior art, single-point sedimentation monitoring or finite element simulation based on homogeneous assumption is mostly adopted, and the fine characterization of the dynamic response of the full section of the roadbed is difficult to realize. Especially for the joint of new and old road surfaces, due to the coupling effect of uneven distribution of filling compactness, fluctuation of groundwater level and traffic load time-course effect, the conventional method often ignores the dynamic evolution mechanism of the weak area, so that the settlement prediction precision is insufficient, and the early warning result and the actual settlement abnormal area have obvious space-time dislocation, so that the timeliness and reliability of engineering risk prevention and control are restricted. Disclosure of Invention The invention mainly aims to provide an intelligent prediction method and system for differential settlement of a highway widening roadbed, and in order to achieve the purposes, the invention provides the intelligent prediction method for differential settlement of the highway widening roadbed, which is applied to an intelligent prediction system for differential settlement of the highway widening roadbed, wherein the system comprises a multi-source data acquisition unit, a creep analysis unit, a coupling correction unit and a settlement prediction unit, and the method comprises the following steps: Based on the sensor, collecting compaction degree distribution data, ground water level change data and traffic load time course curve of the joint of the new road surface and the old road surface in real time, constructing a three-dimensional stress field model with a weak area marking layer, and synchronously outputting stress state parameters of each grid node by the three-dimensional stress field model; The creep analysis unit is used for mapping the weak area marking layer into the filler particle size distribution parameters, dynamically distributing creep parameter weights based on the weak area marking layer coordinates, and generating a creep cloud chart by combining stress state parameters, wherein the creep cloud chart comprises time-varying strain rate distribution strips overlapped with the weak area marking layer; the creep cloud image is matched with grid nodes of a three-dimensional stress field model, strain rate characteristic values corresponding to a weak area marking layer are extracted, a traffic load-shear strength degradation correlation matrix taking a ground water level fluctuation amplitude as a constraint condition is established, and the correlation matrix is dynamically corrected by adopting a double-parameter reduction algorithm based on weak area water content sensitivity; Performing coupling operation on the corrected incidence matrix and stress state parameters, constructing a settlement differential model containing a weak area viscoelastic-plastic constitutive equation set, and generating a settlement distribution map fused with a weak area marking layer space topological structure through space-time convolution; The system synchronously optimizes the spatial resolution and creep parameter weight of the weak area marking layer through triple residual error comparison of a three-dimensional stress field model-creep cloud image-settlement distribution diagram, and outputs a pavement settlement early warning map when the spatial overlap ratio of the weak area marking layer and an actually measured settlement abnormal area reaches a set threshold value. Preferably, the construction of the three-dimensional stress field model is based on multi-source heterogeneous data fusion, the compactness distribution data, the groundwater level change data and the traffic load time course curve are aligned through uniform time stamps, and the data format conversion and the time-space synchronization are realized by adopting weighted superposition; specifically, the compactness data is normalized by taking grid nodes as a reference, the groundwater level data is mapped to the same grid system through spatial interpolation, and a traffic load time course curve is converted into an equivalent node load according to vehicle axle load distribution; the weight distribution is adjusted according to the measurement precision of each data source, wherein the weight coefficient of the compactness data is 0.6, the ground water level data is 0.3, the traffic load data is 0.1, the total weight coefficient is 1,