According to the nonlinear characteristics of slope displacement-time series

the information offered by the time series datum sets can be fully exploited and the trend of slope displacement evolution can be precisely predicted by making real-time forecasting based on phase space reconstruction. The method of local function approximation and the prediction based on Lyapunov exponents are discussed in this paper. The examples suggest that the methods based on phase space reconstruction aremore accurate and more adaptable than the decomposition model of non-stationary time series.