Search

CN-117450994-B - Ground subsidence prediction method, device and medium

CN117450994BCN 117450994 BCN117450994 BCN 117450994BCN-117450994-B

Abstract

The invention discloses a ground subsidence prediction method, a device and a medium, wherein the method comprises the steps of obtaining the ground subsidence time sequence settlement amount of a target point; the method comprises the steps of carrying out EMD decomposition on the settlement of the ground settlement time sequence to obtain IMF components and trend components, utilizing the IMF components and the trend components to establish a prediction model, predicting each component and reconstructing to obtain a prediction result, comparing the prediction precision of settlement when the number of the IMF components is different, selecting the optimal number of IMFs and taking the corresponding prediction result as the final prediction result of a target point. The method can improve the ground subsidence prediction precision and provide basis for analyzing the influence study of ground subsidence on the operation safety of the high-speed rail.

Inventors

  • ZHOU CHAOFAN
  • GONG HUILI
  • CHEN BEIBEI
  • WANG LIN
  • GAO MINGLIANG
  • XU XINYUE

Assignees

  • 首都师范大学

Dates

Publication Date
20260512
Application Date
20230727

Claims (6)

  1. 1. A method of predicting ground subsidence, the method comprising: Acquiring the settlement of the ground settlement time sequence of the target point; EMD decomposition is carried out on the ground settlement time sequence settlement quantity, so that an IMF component and a trend component are obtained; Establishing a prediction model by utilizing the IMF component and the trend component, predicting each component and reconstructing to obtain a prediction result; Comparing the prediction precision of settlement when different IMF component numbers are compared, selecting the optimal IMF number and taking the corresponding prediction result as the final prediction result of the target point; The obtaining the ground subsidence time sequence settlement amount of the target point comprises the following steps: for radar image data of each sub-band, selecting one as a main image and the rest as auxiliary images, and registering the main image and the auxiliary images; selecting a target point from the registered main image and auxiliary image by using an amplitude stability coefficient method; Performing phase analysis on a target point, wherein the phase information contained in the target point is shown in formula (1): (1) Wherein, the Is the phase of the deformation of the earth surface, For the phase of the terrain, For the horizontal phase of the light, In the event of an atmospheric error, Is the noise phase; On the basis of solving the total deformation phase component by using a nonlinear standard model, an external digital elevation model is introduced to remove the terrain phase, orbit data is used to remove the horizon phase, an atmospheric phase model is used to invert and remove the atmospheric phase information, a star topology analysis method is used to invert all deformation phase information of each target point, and the deformation phase information of each target point comprises nonlinear and linear deformation phases; Phase unwrapping the deformation phase information of the target point to obtain the settlement of the ground subsidence time sequence of the target point; EMD (empirical mode decomposition) is carried out on the settlement of the ground settlement time sequence through a formula (2), so that an IMF component and a trend component are obtained: (2) Wherein the method comprises the steps of For the original time series data, using the settlement amount of each high-speed rail along the ground, Is the first The number of IMF components is such that, Is a trend term; The step of establishing a prediction model by using the IMF component and the trend component, predicting each component and reconstructing the component to obtain a prediction result comprises the following steps: selecting the time sequence settlement of the sample points as a data set , Representing the monitoring time, setting the proportion of training set and verification set, and comparing The time series sedimentation amount is constructed as a training set Will be Time series settlement amount is constructed as verification set , Is smaller than Is used for monitoring the time of the test; Initializing for training data sets , Representing the base learner as a decision tree, As a function of the loss, A constant value representing minimizing the loss function; For training data set, M decision trees are established, for Tree-planting, alignment Calculating the negative gradient drop value of the loss function As a model for fitting a decision tree in the next step For the current decision tree model Residual approximation of (2) : ; Obtaining a final model through iteration M times Wherein Consists of M trees; Using a final model And predicting the ground settlement.
  2. 2. The ground subsidence prediction method of claim 1, wherein the smaller the RMSE value and the MAE value, the more accurate the prediction result of the model is represented by comparing the RMSE value and the MAE value with the prediction accuracy of subsidence at different numbers of IMF components.
  3. 3. The ground settlement prediction method according to claim 2, wherein the RMSE value is calculated by the following formula: (3) Wherein the method comprises the steps of The number of sample points is indicated, Representing the first in a time series in a verification dataset The tag value of the individual data is used, Representing the predictive dataset Predicted tag values for the individual data.
  4. 4. A ground settlement prediction method as claimed in claim 3, wherein the RMSE value is calculated by the formula: (4) Wherein the method comprises the steps of The number of sample points is indicated, Representing the first in a time series in a verification dataset The tag value of the individual data is used, Representing the predictive dataset Predicted tag values for the individual data.
  5. 5. A ground settlement prediction apparatus for implementing the method according to any one of claims 1 to 4, the apparatus comprising: the acquisition module is configured to acquire the ground subsidence time sequence settlement amount of the target point; The decomposition module is configured to carry out EMD decomposition on the ground settlement time sequence settlement quantity to obtain an IMF component and a trend component; The prediction module is configured to establish a prediction model by utilizing the IMF component and the trend component, predict each component and reconstruct to obtain a prediction result; the selection module is configured to compare the prediction precision of sedimentation when different IMF component numbers, select the optimal IMF number and take the corresponding prediction result as the final prediction result of the target point.
  6. 6. A readable storage medium, wherein the readable storage medium stores one or more programs, the one or more programs may be executed by one or more processors to implement the method of any of claims 1-4.

Description

Ground subsidence prediction method, device and medium Technical Field The invention belongs to the technical field of geological safety, and particularly relates to a ground subsidence prediction method, a ground subsidence prediction device and a ground subsidence prediction medium. Background The safe operation of the high-speed railway has higher requirements on the stability, deformation degree, track smoothness and the like of roadbeds and bridges. The regional ground subsidence can influence the slope of high-speed railway circuit, and when high-speed railway passed through subsidence funnel district and subsidence area, high-speed railway's operation safety will receive great influence. Jinjin Ji is located in North of Pingyan, regional geology is complex, and water resources are short. To meet the demands of urban construction and sharply increased population, groundwater is overproduced for a long period of time, so that a large area of groundwater overproduction parachute (Gong et al, 2017; guo et al, 2015) is formed in the region. In addition, the Jingjin Ji integrated development, the rapid development of high-density city clusters and the rapid expansion of a high-speed three-dimensional traffic network lead to the formation of a plurality of ground subsidence areas in the area, and the characteristics of cross-area connected sheet distribution are presented, so that the subsidence rate is high (Cui, lei and the like, 2018). The total subsidence area of the ground of Jingjin Ji reaches 1.4X10 4km2, the annual maximum subsidence amount of partial region reaches 160 mm, the serious subsidence area of the ground is mainly distributed in the north and southeast of Beijing plain, in Tianjin Wang Qingtuo region, in Hebei plain, the middle part is fixed, the south and southeast is balanced, cangzhou and south are good, and the subsidence presents the characteristics of cross-region connecting pieces. The regional ground subsidence development affects the operation safety of key linear engineering such as jinghu, jingjin, jin-Bao, shiji high-speed railways, south-water North-transfer and the like, and restricts the regional sustainable development (Gong et al, 2018). The scheme of the fourteenth five-year plan of national economy and society development of the people's republic of China and the scheme of the distant view objective of 2035 indicate that the Jing Ji on the track is basically built, under the background, the prediction of the ground subsidence of the area along the high-speed railway is carried out, the influence of the ground subsidence of the area on the operation safety of the high-speed railway is evaluated, and the scheme has important research value and practical significance on disaster prevention and reduction of other important projects of the Jing Ji under construction. Conventional ground subsidence monitoring techniques, such as leveling, global positioning system (Global Positioning System, GPS) and layering, have certain limitations in large area, high time resolution measurements (Berardino et al, 2002;Galloway D L,1998). The synthetic aperture radar interference (Interferometric synthetic aperture radar, inSAR) measurement technology developed in the 20 th century has the advantages of wide space coverage, high precision, low cost and the like, and can obtain large-area ground subsidence information. Currently, inSAR technology has become a common geodetic method for studying surface deformation, especially permanent scatterer interferometry developed in the technology (PERSISTENT SCATTERER INSAR, PS-InSAR), small baseline set interferometry (Small baseline subset InSAR, SBAS-InSAR), coherent target point analysis technology (Interferometric point TARGET ANALYSIS, IPTA), and the like (Hooper A,2008; gao et al, 2016). These techniques address to some extent the limitations of conventional synthetic aperture radar differential interferometry (D-InSAR) in spatial, temporal decorrelation, atmospheric errors, orbit errors, and terrain error removal. The PS-InSAR technology identifies a point (PERSISTENT SCATTERER, PS) with stable scattering characteristics on a time sequence, and the target point is processed to obtain a reliable surface deformation estimation result, wherein the monitoring precision reaches millimeter level (Ferretti and the like, 2000). In the aspect of ground subsidence prediction research, the method is mainly divided into a deterministic model and a stochastic model. The deterministic model estimates the amount of sedimentation by simulating the physical processes of sedimentation occurrence, development based on physical mechanisms (Ye et al 2005; luo et al 2009; zhu et al 2020). The deterministic model is a mathematical model consisting of an underground water flow model, a soil mechanics model and a coupling model of the underground water flow model and the soil mechanics model, and a large number of physical parameters such as underground water exploi