Displacement prediction of Liangshuijing landslide based on time series additive model
Q X Zhang1, Y P Wu1, 2, G Zh Ou3, X G Fan1, J H Zhou4
COMPUTER MODELLING & NEW TECHNOLOGIES 2014 18(3) 215-223
1 China University of Geosciences, Faculty of Engineering, Wuhan, Hubei, China, 430074
2 Three Gorges Research Center for geo-hazard, Ministry of Education, Wuhan, Hubei, China, 430074
3 Zunyi Normal College, Faculty of Engineering, Zunyi, Guizhou, China, 563002
4 College of Civil Engineering and Architecture, Guilin University of Technology, Guilin, Guangxi, China, 541004
The evolution of landslide displacement is affected by many factors. This paper studied the displacement monitoring data of Liangshuijing Landslide with Factor Analysis Method and found that the dominant factors influencing landslide displacement were in decreasing sequence: cumulative rainfall of anterior two months> rainfall of current month> the average reservoir level of current month>reservoir level fluctuation of current month. The paper selected three typical GPS monitoring points (ZJC09, ZJC11, ZJC13) of Liangshuijing Landslide to forecast their displacements by adopting the time series additive model on basis of the conclusion of previous factor analysis. The accumulative displacement of Liangshuijing Landslide can be divided into trend term and random term. The polynomial fitting was used for trend term displacement prediction. BP neural network model was used for the random displacement prediction. The final calculation results indicated that combination of factor analysis method and time series additive model could generate a reasonable and accurate prediction of landslide.