| description abstract | High slopes are susceptible to weathering, groundwater, rainfall, and other natural factors, resulting in landslides easily occurring on originally stable high slopes. Constructing a reasonable and accurate prediction model for high slopes is equally important to avoid similar accidents, in addition to improving the existing monitoring system. A high slope deformation prediction method based on modified ensemble empirical mode decomposition (MEEMD), permutation entropy (PE), and autoregressive integrated moving average model (ARIMA) is proposed that aims at the characteristics of high slope deformation data, such as multinoise, randomness, and nonstationarity. First, the deformation data are decomposed by MEEMD, and the arrangement entropy value of each component is calculated separately. The components with an entropy value greater than 0.6 are excluded, and those less than 0.6 are reconstructed, to achieve noise elimination; then, the ARIMA model is utilized for the prediction of each component that meets the conditions. Finally, the prediction results of each component are reconstructed to achieve the prediction of the high slope deformation data. The results show that the algorithm can better obtain the actual deformation curve of the high slope and is an immediate method for the prediction of high slope deformation. | |
