A Novel Method for Forecasting Landslide Displacement Based on Phase Space Reconstruction and Support Vector Machine

We presented a novel method for forecasting landslide displacement based on phase space reconstruction and support vector machine.Firstly,based on the chaotic characteristics of landslide displacement time series,mutual information was used to compute the optimal time lag.Then,landslide displacement time series was decomposed into different frequency components through a wavelet transform.We adopted of Cao to compute the optimal embedding dimension of the decomposed series of each component.On the basis of this,phase space reconstruction is performed for each component.Different support vector machines were established to forecast each component.Finally,the predicted results of the components were reconstructed as the final prediction result by wavelet theory.The experimental result show that this method can effectively apply to landslide displacement prediction.