首页 >  , Vol. , Issue () : -

摘要

全文摘要次数: 206 全文下载次数: 125
引用本文:

DOI:

10.11834/jrs.20211297

收稿日期:

2021-05-03

修改日期:

2021-08-13

PDF Free   EndNote   BibTeX
“地面沉降专栏”——基于AM-LSTM网络的北京平原东部地面沉降模拟研究
摘要:

基于传统数值方法构建的地面沉降模拟预测模型需要大量的水文地质数据和实测数据,对于地质条件复杂的地区形变预测难度大。本文基于PS-InSAR技术获取的北京平原东部地区的地面沉降信息,综合考虑不同层位地下水水位对沉降的影响,采用基于注意力机制的长短时记忆网络(AM-LSTM)对不同沉降发育地区典型位置处的地面沉降进行模拟。结果表明:(1)研究区地面沉降空间差异性明显,2010年11月至2016年8月最大沉降速率约153 mm/yr,累计沉降量达到1063 mm,位于朝阳区三间房乡附近;(2)基于AM-LSTM模型的模拟精度优于传统LSTM模型,本次模拟精度最高提升了22%;(3)模拟精度最高的点位于沉降发育较为缓慢的区域,沉降速率处于0~30 mm/yr之间;而模拟精度最差的点位于沉降较为严重的区域,沉降速率大于110 mm/yr;(4)AM-LSTM模型注意力权重表明,第二承压含水层水位对地面沉降贡献最大。研究结果能够为地面沉降防控提供可靠的模型。

Study on land subsidence prediction in the East of Beijing plain based on AM-LSTM Network
Abstract:

The simulation and prediction model of land subsidence based on traditional numerical methods requires a large amount of hydrogeological data and measured data, and it is difficult to predict the deformation in areas with complex geological conditions. In this paper, based on the land subsidence information obtained by PS InSAR Technology in the east of Beijing plain, considering the influence of groundwater level in different layers on the subsidence, the long-term and short-term memory network (AM-LSTM) based on attention mechanism is used to simulate the land subsidence at typical locations in different subsidence areas. The results show that: (1) The spatial difference of land subsidence in the study area is obvious. From October 2010 to August 2016, the maximum subsidence rate was about 153 mm/yr, and the cumulative subsidence reached 1063mm. It is located near Sanjianfang Township, Chaoyang District. (2) The simulation accuracy based on AM-LSTM model is better than that of traditional LSTM model, and the accuracy of this simulation is up to 22%. (3) The point with the highest simulation accuracy is located in the area where the settlement develops slowly, and the subsidence rate is between 0~30 mm / yr. The point with the worst simulation accuracy is located in the area with more serious settlement, and the subsidence rate is greater than 110 mm / yr. (4)The attention weight of AM-LSTM model indicates that the water level of the second confined aquifer contributes the most to land subsidence. The research results can provide a reliable model for the prevention and control of land subsidence.

本文暂时没有被引用!

欢迎关注学报微信

遥感学报交流群 分享按钮