Landslide Prediction Based on Multiple Inducing Factors

Abstract

China is a country with frequent landslide disasters, and the Three Gorges Reservoir area is a landslide disaster-prone area and a serious disaster area. GPS surface displacement monitoring is an important means of landslide stability monitoring. In this paper, we present a novel landslide prediction method based on multiple inducing factors. Firstly, stepwise regression analysis is applied to obtain dominant inducing factors of the landslide. The inducing factors will be processed one by one: the one with significant impact will be retained while the others will be eliminated. Then, each inducing factor will be decomposed by CEEMDAN method, and the components with less influence are eliminated by the gray correlation analysis method. This paper takes the Shuping landslide in the Three Gorges reservoir area as an example. the ELM model is optimized by genetic algorithm, and then the induced factors of optimization are used as input of the model. The experimental results show that the prediction error of the model is relatively small, and the fitting coefficient reaches 0.98. The proposed model has a good effect on landslide prediction.