ACTUAL CASE BASED PREDICTION MODELS FOR LANDSLIDE DEFORMATION

According to the monitored data of Saleshan landslide, three nonlinear prediction models (i.e., exponential model, Verhulst model and grey GM(1,1) model) for the landslide deformation is developed in this paper. The models are based on the studying of combined model with optimum weight and Gauss-Newton method. Then the combined model with optimum weight of the landslide is built. The parameters in each individual model and the combined model are optimized using Gauss-Newton method. By analyzing and comparing, the paper can conclude that the prediction accuracy of the combined model are higher than that of an individual model. The precision for each model after optimizing parameters using Gauss-Newton method is obviously higher than that of the models before using the optimizing parameters. The combined model with the optimizing parameters in all models has the highest precision. So, it is an effective and feasible method for improving landslide prediction to use the combined model with optimum weight and Gauss-Newton method.