Scholarlink

<p>Landslides represent a widespread and highly threatening geological hazard. Real-time monitoring primarily relies on sensors to detect displacement, yet the sole approach of setting displacement thresholds for early warning struggles to capture the dynamic evolution of landslides. It further fails to predict displacement trends, imposing significant limitations on landslide risk assessment.This study proposes an automated landslide displacement stage recognition and short-term risk prediction framework based on the three-stage theory of landslide displacement evolution.The core methodology includes: designing a segmentation algorithm for time-series displacement monitoring data to achieve automatic identification of landslide evolution stages, determining the current disaster development phase, and integrating real-time monitoring data with prior information to assess the current risk. Concurrently, it combines sequential prediction algorithms with an improved tangent angle algorithm model to forecast short-term landslide risks.Through application analysis of real landslide monitoring data, this method accurately achieves automatic classification of landslide deformation stages and risk level assessment. Specifically, it achieves over 92% accuracy in identifying the uniform displacement stage—critical for landslide early warning—while the improved time-series prediction method achieves an RMSE of 0.012 for displacement prediction, enabling effective data forecasting and risk assessment based on the improved tangent angle.The computational method designed in this study, through automated identification of landslide development stages and short-term dynamic displacement prediction, effectively enhances the objectivity, efficiency, and accuracy of early landslide warning systems. It provides a reliable solution for disaster mitigation decision-making in landslide hazards.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Research on landslide evolution stage identification and prediction algorithm based on real-time monitoring data

  • 期刊论文

  • 发表日期：

    2026年2月19日

查看全文
  • Xingchen Wang,
  • Xiaopeng Leng,
  • Lu Jia,
  • Jingping Zhang

摘要

Landslides represent a widespread and highly threatening geological hazard. Real-time monitoring primarily relies on sensors to detect displacement, yet the sole approach of setting displacement thresholds for early warning struggles to capture the dynamic evolution of landslides. It further fails to predict displacement trends, imposing significant limitations on landslide risk assessment.This study proposes an automated landslide displacement stage recognition and short-term risk prediction framework based on the three-stage theory of landslide displacement evolution.The core methodology includes: designing a segmentation algorithm for time-series displacement monitoring data to achieve automatic identification of landslide evolution stages, determining the current disaster development phase, and integrating real-time monitoring data with prior information to assess the current risk. Concurrently, it combines sequential prediction algorithms with an improved tangent angle algorithm model to forecast short-term landslide risks.Through application analysis of real landslide monitoring data, this method accurately achieves automatic classification of landslide deformation stages and risk level assessment. Specifically, it achieves over 92% accuracy in identifying the uniform displacement stage—critical for landslide early warning—while the improved time-series prediction method achieves an RMSE of 0.012 for displacement prediction, enabling effective data forecasting and risk assessment based on the improved tangent angle.The computational method designed in this study, through automated identification of landslide development stages and short-term dynamic displacement prediction, effectively enhances the objectivity, efficiency, and accuracy of early landslide warning systems. It provides a reliable solution for disaster mitigation decision-making in landslide hazards.