基于GNSS与分布式光纤传感的滑坡实时形变监测与早期预警研究

Real-Time Deformation Monitoring and Early Warning of Landslides Using GNSS and Distributed Fiber Optic Sensing

摘要: 滑坡灾害的早期识别与实时预警是地质灾害防治领域的关键重点。传统监测方法受限于空间分辨率与实时性，难以有效捕捉滑坡多尺度变形过程。本文以云南省小龙洞滑坡为研究对象，构建了一套融合全球导航卫星系统(GNSS)与分布式光纤传感技术(Fiber Bragg Grating, FBG)的多尺度实时形变监测与预警系统。通过建立“区域–主体–关键点”三级监测体系，实现了从宏观地表位移至微观应变响应的全断面协同感知。本文基于多源异构数据的实时采集与高精度融合技术，根据自适应卡尔曼滤波算法实现了GNSS绝对位移与FBG应变数据的时空统一与联合建模，并结合长短期记忆网络(LSTM)与多指标阈值机制，建立了位移速率、应变突变与时空关联特征协同的多级预警模型。系统实际运行表明，其可有效识别区域背景位移、主体蠕变与局部应变异常，预警响应延迟小于3分钟，历史预警准确率达93%，灾害识别时效性提升72%。现场实测结果验证GNSS与光纤传感融合技术在滑坡灾害早期预警中的有效性，为复杂地质环境下的多尺度监测提供了可靠的技术支撑。

Abstract: The early identification and real-time warning of landslide hazards represent critical priorities in the field of geological disaster prevention and mitigation. Conventional monitoring methods are often limited by spatial resolution and real-time performance, making it difficult to effectively capture the multi-scale deformation processes of landslides. This study takes the Xiaolongdong landslide in Yunnan Province as a case study and develops an integrated multi-scale real-time deformation monitoring and early warning system that combines Global Navigation Satellite System (GNSS) and Fiber Bragg Grating (FBG)-based distributed fiber optic sensing technology. By establishing a three-level monitoring system, the system achieves coordinated full-section perception from macroscopic surface displacement to microscopic strain response. Based on real-time acquisition and high-precision fusion of multi-source heterogeneous data, spatiotemporal unification and joint modeling of GNSS absolute displacement and FBG strain data are accomplished using an adaptive Kalman filtering algorithm. Furthermore, a multi-level early warning model is established by integrating Long Short-Term Memory (LSTM) networks and a multi-index threshold mechanism, which incorporates displacement rate, abrupt strain changes, and spatiotemporal correlation features. System operation results demonstrate that the system can effectively identify regional background displacement, main body creep, and local strain anomalies, with a warning response delay of less than 3 minutes, a historical warning accuracy rate of 93%, and a 72% improvement in disaster identification timeliness. Field measurements verify the effectiveness of the integrated GNSS and fiber optic sensing technology in early landslide warning, providing reliable technical support for multi-scale monitoring in complex geological environments.