PREDICTION OF LANDSLIDE DISPLACEMENT TIME SERIES BASED ON SUPPORT VECTOR REGRESSION MACHINE WITH ARTIFICIAL BEE COLONY ALGORITHM

A method of landslide displacement prediction based on time series analysis model is proposed. It combines artificial bee colony algorithm(ABC)with support vector regression machine(SVR). The existing problems of landslide displacement prediction methods are summarized. We select Baishuihe landslide in The Gorges Reservoir area as the research object. We study the influence of landslide displacement, rainfall, reservoir water level and other factors on the change of landslide displacement with time. Firstly, the landslide displacement is decomposed into a trend term and a periodic term by time series addition model and moving average method. We use the polynomial least square method to fit and predict the trend term of landslide displacement. Then we use artificial bee colony support vector machine model to train and predict the periodic term of landslide displacement. In this paper, seven factors affecting the displacement of periodic terms are selected for the analysis. We use the grey system correlation analysis method to calculate the correlation degree between each factor and the displacement of the same period term. The total displacement prediction value of landslide is the sum of trend and periodic displacement prediction values. Compared with BP neural network and PSO-SVR model, this method has higher accuracy in landslide displacement prediction, and has better application prospects in disaster prevention and mitigation.