Analysis Of Landslide Prediction By Chaotic Nonlinear Time Series

Slope is an unavoidable geological environment to human beings, accompanied with human engineering activities always. On the one hand, people are trying hard to convert and consolidate slope in order to benefit human beings, on the other hand, slope which are disturbed by human activities and outer surroundings will collapse, which bring about hazards to the security of human life and construction activities. Because slope is a complicated system, all sorts of parameters itself are uncertain and random. During the process of evolution, they exchange substance, energy and information with outside continuously and their activities manifest complicated nonlinear actions. Therefore, the dissertation adopt one of nonlinear theory, chaos theory, to analyze the slope.In the dissertation, we summarize the analytic condition of slope prediction all over the world and introduce chaos theory and chaotic characters of the slope systematically. The basis of slope prediction with chaos theory is the phase space reconstruction of time series. The paper elaborate the method of the phase space reconstruction and summarize the methods of time delay and embedding dimension. The chaotic invariants of measured time series of Maoping landslide such as correlation dimension, the Lyapunov exponent and entropy are calculated. The results show that the monitoring data obtained in Maoping landslide is a chaotic time series.The theory of prediction is discussed and the several prediction methods that are often used at present are introduced, such as local constant prediction, local linear prediction and the largest Lyapunov exponent prediction. For the Lyapunov exponent is the character of chaotic time series, the largest Lyapunov exponent prediction is more accurate than others.Finally, prediction methods to multivariate chaotic time series are studied,based on the above univariate prediction methods. Under the same circumstances but half of the data length, numerical emulation calculations verify that multivariate time series prediction better than univariate time series both with local prediction methods and global ones. It has actual engineering meanings for landslide prediction.