基于VMD-Kalman-GM组合模型的滑坡位移预测

Landslide Displacement Prediction Based on a Combined VMD-Kalman-GM Model

滑坡位移的预测与分析能为滑坡灾害的预警提供重要数据支持作用,针对受降雨影响存在波动发展的滑坡位移序列,为了提高降雨对滑坡位移影响的预测效果,文章建立了一种变分模态分解(VMD)算法和卡尔曼滤波(Kalman)、改进灰色模型(GM)组合预测的方法,文章基于变分模态分解算法将滑坡地表监测位移序列分解不同频率分量,经过时序组合得到波动值和趋势值,在确定波动值与降雨数值时滞相关性的条件下,提出了一种考虑降雨数值的变化趋势的卡尔曼滤波预测模型,建立降雨时间滞后影响下的卡尔曼滤波预测模型,利用该模型进行滑坡位移波动值的动态预测,同时建立动态灰色预测模型预测趋势值,最后波动值和趋势值合成得到滑坡预测数值,建立了VMD-Kalman-GM组合预测模型。以中国三峡库区八字门滑坡监测数据为例,将预测结果与实测值进行比较,验证了该方法的可行性和准确性,为滑坡位移的预测提供了一种新的方法。

In order to improve the prediction effect of rainfall on landslide displacements,the article establishes a combined prediction method of variational modal decomposition(VMD)algorithm and improved Kalman filter(Kalman)and improved grey model(GM).The article decomposes the landslide surface monitoring displacement series into different frequency components based on the variational modal decomposition algorithm,and obtains the fluctuation and trend values after the time series combination.The VMD-Kalman-GM combined prediction model was established by combining the dynamic prediction of the landslide displacement with the dynamic grey prediction model to predict the trend value,and finally synthesizing the fluctuation value and the trend value to obtain the landslide prediction value.The prediction results were compared with the actual measured values using the monitoring data of the Bazimen landslide in the Three Gorges reservoir area of China as an example,verifying the feasibility and accuracy of the method and providing a new method for the prediction of landslide displacements.