A Novel Changeable Sliding Window Method for Predicting Horizontal Displacement of Dam Foundation

The prediction of horizontal displacement is significant important to the safety-control in dam. A number of regression analysis-based methods have been proposed. However, most of the methods have unsatisfied performance on prediction accuracy in case of the sample size is unknown. In this paper, a method based on changeable sliding window is proposed to predict horizontal displacement of dam foundation dynamically, which the optimal fitting model and samples can be selected through a way of multiple iterations. The experiment results show that the fair changing pattern can be discovered by the proposed method, and it is also suitable for the prediction in other area.