基于时间序列分析的公路降雨型滑坡位移预测

基于公路降雨型滑坡位移远程实时监测数据，从分析降雨型滑坡阶跃型变形规律出发，将滑坡总位移分解为受不利地质条件影响的趋势项位移以及受降雨影响的周期项位移，采用时间序列分析方法将两者叠加构建降雨型滑坡位移预测模型。对滑坡趋势项位移采用灰色等维新息模型进行预测，对周期项位移采用自回归模型进行预测，将两者预测值组合得到滑坡总位移预测值。以湖南娄新高速公路某滑坡实时监测预警为例，采用该模型对比分析了滑坡实测与预测时间位移曲线，表明该模型预测效果较好，能体现降雨型滑坡位移变化趋势，具有较好的应用前景。

Displacement Prediction of Rainfall-induced Highway Landslides Based on Time Series Analysis

Based on the real data from remote displacement monitoring of rainfall-induced highway landslide and the analysis of the stepwise deformation law thereof, the displacement prediction model of rainfall-induced landslide was established in such manner that the total displacement was decomposed into trend component influenced by unfavorable geological conditions and periodic component influenced by rainfall, both of which were superposed by means of time series analysis. The trend component was predicted with grey equal dimension new information model, while the periodic component with auto-regressive model, then the predicted value of total displacement may be obtained by the superposition of the two predicted values mentioned above. With this model, and the real monitoring for early warning on a landslide along Loudi-Xinhua Highway in Hunan taken as an example, the time-displacement curves measured and predicted were compared and analyzed. It is found that this model is efficient, and it can show the variation trend of displacement of rainfall-induced landslide and has a better application prospect.