Abstract:To overcome the deficiency of the single forecasting model, an EEMD-PSOGSVM prediction model of foundation pit displacement is proposed, based on chaotic time series. The EEMD is adapted to decompose the time series, then phase space reconstruction technique is used to reconstruct the sample. The PSOGSVM model is then applied to predict. A comparative study of some deep foundation pit displacement is made by using the GM (1, 1), SVM and wavelet neural network optimized by genetic algorithm models, respectively. The results show that the predictive accuracy of this method is better and more stable, and that it can be effectively applied into the prediction of foundation pit displacement.