Prediction of surface deformation time series in closed mines based on LSTM and optimization algorithms

Underground coal mining impacts surface deformation not only during mining operations but also after mine closures, potentially causing prolonged subsidence or uplift. Long-term predictions of surface deformation in closed mining areas remain challenging, as traditional statistical methods have limi...

Full description

Saved in:
Bibliographic Details
Main Authors: Hu Caixiong, Zhang Lili, Li Haoran, Zhang Yaowen, Yao Yunsheng
Format: Article
Language:English
Published: De Gruyter 2025-06-01
Series:Open Geosciences
Subjects:
Online Access:https://doi.org/10.1515/geo-2025-0811
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850124649259597824
author Hu Caixiong
Zhang Lili
Li Haoran
Zhang Yaowen
Yao Yunsheng
author_facet Hu Caixiong
Zhang Lili
Li Haoran
Zhang Yaowen
Yao Yunsheng
author_sort Hu Caixiong
collection DOAJ
description Underground coal mining impacts surface deformation not only during mining operations but also after mine closures, potentially causing prolonged subsidence or uplift. Long-term predictions of surface deformation in closed mining areas remain challenging, as traditional statistical methods have limited capability in modeling complex spatiotemporal deformation patterns. To address this, this study utilized the small base line subsets interferometric synthetic aperture radar technique to derive deformation results in the line-of-sight direction for the study area. A long short-term memory (LSTM) neural network combined with the gray wolf optimizer (GWO) algorithm was introduced to improve prediction accuracy. Experimental results showed that the LSTM model achieved a correlation coefficient of 0.93 with the observed data, while the GWO–LSTM model further enhanced prediction accuracy with a correlation coefficient of 0.99, demonstrating its potential for complex deformation prediction. The findings provide a scientific basis for surface deformation prediction and risk assessment in closed mining areas and have practical significance for disaster warning and decision-making in the region. As synthetic aperture radar data availability continue to grow, this method holds promise for broader applications in other geological environments.
format Article
id doaj-art-1ad0c5618e9f4a2aa2e610e2c76b929d
institution OA Journals
issn 2391-5447
language English
publishDate 2025-06-01
publisher De Gruyter
record_format Article
series Open Geosciences
spelling doaj-art-1ad0c5618e9f4a2aa2e610e2c76b929d2025-08-20T02:34:16ZengDe GruyterOpen Geosciences2391-54472025-06-0117117054110.1515/geo-2025-0811Prediction of surface deformation time series in closed mines based on LSTM and optimization algorithmsHu Caixiong0Zhang Lili1Li Haoran2Zhang Yaowen3Yao Yunsheng4Institute of Disaster Prevention, Sanhe, 065201, Hebei, ChinaInstitute of Disaster Prevention, Sanhe, 065201, Hebei, ChinaInstitute of Disaster Prevention, Sanhe, 065201, Hebei, ChinaInstitute of Disaster Prevention, Sanhe, 065201, Hebei, ChinaInstitute of Disaster Prevention, Sanhe, 065201, Hebei, ChinaUnderground coal mining impacts surface deformation not only during mining operations but also after mine closures, potentially causing prolonged subsidence or uplift. Long-term predictions of surface deformation in closed mining areas remain challenging, as traditional statistical methods have limited capability in modeling complex spatiotemporal deformation patterns. To address this, this study utilized the small base line subsets interferometric synthetic aperture radar technique to derive deformation results in the line-of-sight direction for the study area. A long short-term memory (LSTM) neural network combined with the gray wolf optimizer (GWO) algorithm was introduced to improve prediction accuracy. Experimental results showed that the LSTM model achieved a correlation coefficient of 0.93 with the observed data, while the GWO–LSTM model further enhanced prediction accuracy with a correlation coefficient of 0.99, demonstrating its potential for complex deformation prediction. The findings provide a scientific basis for surface deformation prediction and risk assessment in closed mining areas and have practical significance for disaster warning and decision-making in the region. As synthetic aperture radar data availability continue to grow, this method holds promise for broader applications in other geological environments.https://doi.org/10.1515/geo-2025-0811sbas-insardeep learningclosed minelstmprediction
spellingShingle Hu Caixiong
Zhang Lili
Li Haoran
Zhang Yaowen
Yao Yunsheng
Prediction of surface deformation time series in closed mines based on LSTM and optimization algorithms
Open Geosciences
sbas-insar
deep learning
closed mine
lstm
prediction
title Prediction of surface deformation time series in closed mines based on LSTM and optimization algorithms
title_full Prediction of surface deformation time series in closed mines based on LSTM and optimization algorithms
title_fullStr Prediction of surface deformation time series in closed mines based on LSTM and optimization algorithms
title_full_unstemmed Prediction of surface deformation time series in closed mines based on LSTM and optimization algorithms
title_short Prediction of surface deformation time series in closed mines based on LSTM and optimization algorithms
title_sort prediction of surface deformation time series in closed mines based on lstm and optimization algorithms
topic sbas-insar
deep learning
closed mine
lstm
prediction
url https://doi.org/10.1515/geo-2025-0811
work_keys_str_mv AT hucaixiong predictionofsurfacedeformationtimeseriesinclosedminesbasedonlstmandoptimizationalgorithms
AT zhanglili predictionofsurfacedeformationtimeseriesinclosedminesbasedonlstmandoptimizationalgorithms
AT lihaoran predictionofsurfacedeformationtimeseriesinclosedminesbasedonlstmandoptimizationalgorithms
AT zhangyaowen predictionofsurfacedeformationtimeseriesinclosedminesbasedonlstmandoptimizationalgorithms
AT yaoyunsheng predictionofsurfacedeformationtimeseriesinclosedminesbasedonlstmandoptimizationalgorithms