| 摘要: |
| 借助时序InSAR监测技术可为土石坝表面变形全覆盖监测提供支撑，对弥补传统地面单测点监测的不足具有重要意义。针对时序InSAR所获得的海量监测数据分析困难，提出了土石坝表面变形态势的聚类分析和异常变形区域识别方法。首先依据InSAR监测数据所表征的大坝表面变形规律，采用层次聚类算法对坝体表面进行分区；再利用云模型的逆向云发生器将InSAR相干点的变形序列转化为云参数，概化各分区的变形特征；最后借助局部异常因子量化分区内各相干点的异常程度，以识别出异常变形区域。工程实例表明，所提出的聚类分析方法可以对海量InSAR监测数据进行高效分区，能够有效识别土石坝异常变形区域，为提升时序InSAR监测技术应用于土石坝沉降变形态势分析提供了实用的数据分析方法。 |
| 关键词: 合成孔径雷达干涉 大坝安全监测 聚类分析 云模型 局部异常因子 |
| DOI： |
| 分类号: |
| 基金项目:国家重点研发计划项目(2021YFB3900603),中央级公益性科研院所基本科研业务费专项资金项目(Y724002), 中央级公益性科研院所基本科研业务费专项资金项目(Y723005)，南京水利科学研究院研究生学位论文发展基金（Yy724008）。 |
|
| Clustering analysis of earth-rock dam settlement deformation characteristics based on Time Series InSAR monitoring |
|
LI Ziyang, WANG Wen, LOU Benxing, LI Qiang, LI Hanman
|
|
Nanjing Hydraulic Research Institute
|
| Abstract: |
| Time series InSAR monitoring technology can provide comprehensive support for full-coverage surface deformation monitoring of earth-rock dams, significantly addressing the limitations of traditional ground-based single-point monitoring. In response to the challenges in analyzing the massive dataset obtained from time series InSAR monitoring, this paper proposes a clustering analysis method for identifying abnormal deformation areas on the earth-rock dam surface. Firstly, according to the deformation patterns of the dam surface represented by InSAR monitoring data, hierarchical clustering is applied to divide the dam body into different zones. Then, the inverse cloud generator of the cloud model is utilized to transform the deformation sequences of InSAR coherent points into cloud parameters, generalizing the deformation characteristics of each zone. Finally, local anomaly factors are used to quantify the degree of abnormality of each coherent point within zones to identify areas with abnormal deformations. Engineering case studies demonstrate that the proposed clustering analysis method can efficiently partition large-scale InSAR monitoring datasets and effectively identify areas of abnormal deformation on earth-rock dams, providing a practical data analysis method to enhance the application of time series InSAR monitoring technology in analyzing the settlement characteristics of dam surfaces. |
| Key words: synthetic aperture radar interferometry dam safety monitoring clustering analysis cloud model local outlier factor |
