1. 基于混沌时间序列的黄土滑坡变形预测方法及应用.
- Author
-
王 利, 岳 聪, 舒 宝, 张耀辉, 许 豪, and 义 琛
- Subjects
- *
TIME series analysis , *LYAPUNOV exponents , *CHAOS theory , *MEASUREMENT errors , *LANDSLIDES , *DEFORMATION of surfaces - Abstract
Due to the existence of observation noise such as multi-path error, the accuracy of deformation prediction results are affected by using the data series of GNSS deformation monitoring. In order to examine the influence of measurement error on the deformation prediction results, the GNSS derived surface displacement time series of Miaodian landslide in Jingyang Area of Shaanxi, and those after noise suppression in combination with chaos theory were analyzed. Firstly, the mutual information method was used to determine the time delay of the surface displacement time series, and the Cao method was used to determine the embedding dimension to obtain the phase space reconstruction parameters. Secondly, the maximum Lyapunov exponent method was used to identify the chaotic characteristics of the two surface displacement time series. Finally, the weighted first-order local prediction method, the largest Lyapunov exponent prediction method, and the BP neural network prediction method were used to predict the landslide surface displacements. The results show that the GNSS landslide surface displacement time series and the time series after noise suppression have chaotic characteristics. The BP neural network prediction method has good prediction performance with an MAE of 0.4 mm and an MRE of 11.9%. After S-transform noise suppression, the MAE and MRE are 0.1 mm and 4.12%, respectively. Compared with the original time series, the prediction performance has been significantly improved after noise suppression. [ABSTRACT FROM AUTHOR]
- Published
- 2021
- Full Text
- View/download PDF