Landslide hazard early warning method for rock slopes using a hybrid LSTM-SARIMA data-driven model

Yongxin Dai, Zijian Li and Jingbiao Lu

PLOS ONE, 2025, vol. 20, issue 5, 1-29

Abstract: Rock slope landslides are characterized by their sudden onset and significant destructive power, posing a major threat to human life as well as the safety of equipment and infrastructure.Currently, research on landslide early hazard warning has largely focused on individual components, such as monitoring data analysis or studies on influencing mechanisms. However, landslide early hazard warning is a complex, multi-stage technical system where each stage is closely interlinked, and focusing solely on a single component cannot fulfill the objectives of effective monitoring and warning. This paper proposes a comprehensive technical system for landslide early hazard warning in open-pit mine slopes, encompassing the full process of monitoring data acquisition and processing, analysis of influencing mechanisms, intelligent algorithm-based prediction, and the construction of early hazard warning indicators. Each stage of the early hazard warning process is systematically researched and summarized.First, the combination of sliding average and wavelet noise reduction is utilized to perform global denoising and local focus noise reduction on the original monitoring data, and the signal-to-noise ratios after two rounds of noise reduction are 36 and 44, respectively, which indicates a good noise reduction effect. The Hodrick-Prescott (HP) filter is used to split the slope displacements into components, the Long Short-Term Memory (LSTM)–Seasonal Autoregressive Integrated Moving Average (SARIMA) hybrid model is proposed to predict the slope of the trend term of displacements and period term of displacements, and the prediction accuracy of the LSTM–SARIMA hybrid model reaches 96%. The excellence of the hybrid-driven model was determined by introducing five data-driven models, a Support Vector Machine (SVM), a Random Forest (RF),eXtreme Gradient Boosting (XGBoost),Recurrent Neural Network(RNN) and Light Gradient Boosting Machine(LightGBM), for comparison.Finally, the improved tangent angle of the T-t curve is employed as the landslide warning criterion, enabling accurate prediction of landslide events in an open-pit mine in East China. The successful application of this system demonstrates that the comprehensive warning framework proposed in this study can accurately predict the occurrence of rock slope landslides.

Date: 2025
References: View complete reference list from CitEc
Citations:

Downloads: (external link)
https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0323650 (text/html)
https://journals.plos.org/plosone/article/file?id= ... 23650&type=printable (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:plo:pone00:0323650

DOI: 10.1371/journal.pone.0323650

Access Statistics for this article

More articles in PLOS ONE from Public Library of Science
Bibliographic data for series maintained by plosone ().