| 以下仅列出文章摘要、提纲简介，如需获取全文阅读权限，或原创定制、长期合作，请随时联系。 微信QQ:312050216 |
 |
 |
| 基于北斗GNSS路堑高边坡变形预测和预警机制研究
摘要
本文依托某高速公路路堑高边坡工程，结合北斗GNSS系统大量的监测数据，运用基于深度学习的FOS-ELM算法建立高边坡变形预测模型，并根据相关规范中边坡性安全系数，确定了高边坡正常工况下预警阈值和降雨条件下短期变形预警阈值，最终形成了高边坡变形的预警机制。通过本文的研究分析，获得了如下主要成果：
（1）通过北斗GNSS系统收集高边坡位移原始数据，运用载波相位差分定位方法解算位移量，实现边坡高精度位移监测。通过与传统人工测量结果对比，北斗GNSS变形监测精度可以达到人工测量的毫米级精度，变形整体发展趋势与人工结果总体一致。
（2）采用基于深度学习的FOS-ELM算法建立了高边坡变形预测模型，实现了监测点的单步预测和多步超前预测。单步预测结果与实际监测值基本一致，多步超前预测效果随超前预测时间增加而显著降低，超前6小时预测结果与实测值较为接近。提出了考虑降雨作用的高边坡变形预测模型中，耦合降雨量可以完成单步变形预测，且预测结果在可接受范围内。
（3）基于有限元分析软件确定各边坡临界状态对应的贯通塑性区，并通过强度折减方法确定各监测点临界水平位移与安全系数关系，依据现有规范的边坡稳定状态划分方法可以确定各监测点在不同安全等级下的长期变形预警指标，从而结合高边坡变形预测实现了边坡预警机制研究。
关键词：边坡；北斗GNSS；变形；预测；滑坡预警
Abstract
Relying on the high cutting slope engineering of an expressway,this paper combined with beidou GNSS system a lot of monitoring data, using FOS - ELM algorithm based on depth of learning high slope deformation prediction model is established,and according to the relevant specification slope safety factor, determines the high slope under normal working condition of early warning threshold and short-term deformation under the condition of rainfall warning threshold, finally formed the high slope deformation of the early warning mechanism. Through the research and analysis of this paper, the following main results are obtained:
(1) The original displacement data of the high slope is collected by the Beidou GNSS system, and the carrier phase differential positioning method is used to calculate thedisplacement, so as to realize the high-precision displacement monitoring of the slope. Compared with the traditional manual measurement results,the deformation monitoring accuracy of Beidou GNSS can reach the millimeter-level accuracy of manual measurement, and the overall development trend of deformation is consistent with the artificial results.
(2) The Fos-ELM algorithm based on deep learning was used to establish the high slope deformation prediction model,and the single-step prediction and multi-step advance prediction of the monitoring points were realized. The results of single step prediction are basically consistent with the actual monitoring values,and the effect of multi-step advance prediction decreases significantly with the increase of advance prediction time, and the prediction results of 6 hours ahead are close to the measured values. In the prediction model of high slope deformation considering the effect ofrainfall, coupled with rainfall,the single-step deformation prediction can be completed, and the prediction results are in an acceptable range.
(3) the critical state of slope is determined based on the finite element analysis software corresponding well versed in the plastic zone, and through the strength reduction method to determine the critical level of the monitoring displacement and safety coefficient,on the basis of existing specifications of slope stable state method can determine the long-term deformation under different security levels of the monitoring early warning indicators, and combined with high slope deformation prediction to achieve the study of early warning mechanism of slope.
Key words: Slope；Beidou GNSS; Deformation; Prediction; Landslide warn
|
