Stability Analysis and Deformation Prediction of Open-Pit Mine Slopes Based on InSAR and COMSOL
Rapid and accurate analysis of surface deformation characteristics and accurate prediction of deformation trends in open-pit mines is an important guarantee for promoting green and safe production in mines.Aiming at the problems of the current deformation monitoring techniques,such as low temporal and spatial sampling rate,high cost,and difficult to deter-mine the parameters of prediction model,an integrated method of slope stability analysis and deformation prediction based on short baseline subset interferometry(SBAS-InSAR)technique and COMSOL finite element simulation was proposed by taking East Anshan Open-pit Iron Mine as the engineering background.Firstly,62 views of Sentinel-1A up-orbit SAR data acquired from May 2018 to June 2020 were processed by SBAS-InSAR technology.The time series of surface deformation in the region within 2 years could be acquired and the temporal and spatial evolution characteristics of the deformation,and the spatial and temporal evolution characteristics of the deformation were analyzed.Then,COMSOL software was used to simulate the slope sta-bility condition of a typical subsidence area under the influence of external heavy rainfall.The damage cracking law and de-formation mechanism of the slope were explored.Based on this,the particle swarm algorithm(PSO)was adopted to optimize the long-short-term time memory(LSTM)network to build an optimal model for deformation time series prediction and to carry out deformation time series prediction within the typical settlement area.The average absolute error and root mean square error were introduced as the evaluation indexes of prediction accuracy.The results show that the subsidence in the western part of the mining area is relatively serious,and the average annual subsidence rate is as high as 47.8 mm/a.There is a significant corre-lation between the deformation rate and the quantity of local rainfall.Compared with the traditional deformation prediction mod-el,the two errors of PSO-LSTM model are reduced by at least 14%and 36%respectively.It can effectively reflect the fluctua-tion trend of surface deformation in the mining area,which provides a potential approach for early warning of landslides.
NETL
NSTL
万方数据
