Prediction Of Landslide Displacemeng Based On Chaos And Ann Theory

The evolution of landslide has tremendous influence to human life. Predicting the evolution of landslide accurately has important significance. The landslide system is a complex nonlinear dynamic system under the influence of various factors such as geological conditions, underground water, and human activity. Its evolution rule is extremely complicated, which follows some objective laws apparently and seems to be random at the same time. It causes that predicting the evolution of landslide accurately is so difficult. The chaos theory and artificial neural networks theory are introduced into the model of landslide prediction and a new method of landslide prediction is put forward in this thesis.The landslide system is controlled by many factors which cooperate and complete with each other. When force of each factor is equal, the evolution of landslide shows extremely intricate chaos character. The time series of landslide displacement is used for research in this thesis because it is obtained easily in practice project. The main content is as follows:Based on the research of basic chaos theory, reconstruction of phase space which is suited to the time series of landslide displacement is put forward. And the time series of Dayantang landslide displacement is reconstructed using this method. The embedding dimension and delay time of phase space are found. Then the chaos character of landslide is researched in the reconstructed phase space. In this research, the correlation dimension and the maximum lyapunov exponent of Dayantang landslide are computed, which prove Dayantang landslide has the characteristic of chaos.Based on the study of network training and network design of BP artificial neural networks, a new global predicting model by BP network is built, which fits for the time series of landslide displacement. The phase points in the reconstructed phase space of Dayantang landslide are fitted and the landslide displacements are predicted by this model. The predicted result is better than the result by the weighted zero-rank local method.In a word, the model based on chaos theory and artificial neural networks theory is able to reflect the characteristic of landslide evolution and predict the landslide displacement by the limited monitoring data fully. It is a new predicting model which has research value.