Scholarlink

<p>Early warning of slope failure is an effective approach for reducing casualties and property losses associated with landslides. This study proposes a coupled slope early-warning model that integrates a time-dependent degradation model with data-driven monitoring analysis. First, a time-dependent degradation model is calibrated using a portion of the monitoring data to generate a training dataset representing the expected displacement evolution. Then, kernel density estimation is employed to construct a residual probability distribution function describing the differences between the training dataset and the observed monitoring data. The calibrated time-dependent degradation model is then used to predict future slope displacement, forming a predicted dataset. The residuals between the predicted displacements and the actual monitoring measurements are continuously evaluated. Deviations of these residuals from the established probability distribution indicate potential progressive slope destabilization. Multi-level early-warning thresholds are defined based on various confidence levels of the residual distribution. The proposed model is validated using a landslide case from the Huolinhe open-pit mine (China). When confidence levels of 0.05, 0.01, and 0.001 are adopted corresponding to moderate anomaly, severe anomaly, and extreme anomaly, the system issued warnings 2204, 2056, and 2042 min, respectively, before slope failure occurred. These results demonstrate that the proposed classification-based early-warning model can provide timely alerts prior to slope failure. The model enables multi-level risk identification and supports proactive decision-making for slope hazard mitigation.</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

An Early-Warning Model for Slope Failure Integrating Time-Dependent Degradation and Residual Distribution Analysis

  • 期刊论文

  • 发表日期：

    2026年5月26日

查看全文
  • Zhiguo Li,
  • Tao Xu,
  • Lichun Zhao,
  • Yongjie Liu,
  • Michael J. Heap,
  • Yongchao Xu,
  • David Amitrano,
  • Xiaolong Qiao

摘要

Early warning of slope failure is an effective approach for reducing casualties and property losses associated with landslides. This study proposes a coupled slope early-warning model that integrates a time-dependent degradation model with data-driven monitoring analysis. First, a time-dependent degradation model is calibrated using a portion of the monitoring data to generate a training dataset representing the expected displacement evolution. Then, kernel density estimation is employed to construct a residual probability distribution function describing the differences between the training dataset and the observed monitoring data. The calibrated time-dependent degradation model is then used to predict future slope displacement, forming a predicted dataset. The residuals between the predicted displacements and the actual monitoring measurements are continuously evaluated. Deviations of these residuals from the established probability distribution indicate potential progressive slope destabilization. Multi-level early-warning thresholds are defined based on various confidence levels of the residual distribution. The proposed model is validated using a landslide case from the Huolinhe open-pit mine (China). When confidence levels of 0.05, 0.01, and 0.001 are adopted corresponding to moderate anomaly, severe anomaly, and extreme anomaly, the system issued warnings 2204, 2056, and 2042 min, respectively, before slope failure occurred. These results demonstrate that the proposed classification-based early-warning model can provide timely alerts prior to slope failure. The model enables multi-level risk identification and supports proactive decision-making for slope hazard mitigation.