海洋测绘2024，Vol.44Issue(3)：47-52,6.DOI:10.3969/j.issn.1671-3044.2024.03.010
门控循环神经网络的时序PS-InSAR地面沉降预测
The prediction of the time-series PS-InSAR ground subsidence based on gated recurrent neural network
摘要
Abstract
To prevent damage to infrastructure caused by ground subsidence caused by land reclamation at Hong Kong International Airport,China,this paper used the Persistent Scatterer Interferometric Synthetic Aperture Radar(PS-InSAR)technique to obtain ground subsidence data for 2016-2020 at Hong Kong International Airport,China.We used the Small Baseline Subset Interferometric Synthetic Aperture Radar(SBAS-InSAR)technology and level data to verify the results of PS-InSAR.Moreover,Gated Recurrent Unit(GRU)neural network was introduced to construct a stacked GRU ground subsidence prediction model,and the future ground subsidence of Hong Kong International Airport in China was predicted in time-series,and compared with Support Vector Machine(SVM)and Multilayer Perceptron(MLP)neural network.The results showed that the spatial distribution of ground subsidence at Hong Kong International Airport,China,from 2016 to 2020 was uneven,the accumulated subsidence gradually increased,and the accumulated subsidence in the vertical direction in December 2020 reached 106 mm.The constructed stacked GRU neural network ground subsidence method was more accurate than SVM and MLP,and the maximum accumulated ground subsidence at Hong Kong International Airport,China,in July 2021 reached 111.8 mm.The ground subsidence time series prediction model proposed in this paper can be used as an effective method to predict ground subsidence and provide key technical support for early warning of ground subsidence.关键词
永久散射体合成孔径雷达干涉测量/地面沉降/时序预测/门控循环神经网络/填海造陆Key words
PS-InSAR/ground subsidence/time-series prediction/GRU neural network/land reclamation分类
天文与地球科学引用本文复制引用
火天宝,何毅,姚圣,张立峰,张清..门控循环神经网络的时序PS-InSAR地面沉降预测[J].海洋测绘,2024,44(3):47-52,6.基金项目
国家自然科学基金(42201459) （42201459）
甘肃省教育厅:青年博士基金(2022QB-058) （2022QB-058）
甘肃省自然科学基金(20JR10RA249). （20JR10RA249）
