A Dynamic Modeling Method of Dam Safety Early-Warning Model

An early-warning model of dam safety base on Dynamic FCM-LSSVM was supported to solve the traditional model failure over time. The training set was divided into subsets with different cluster centers method by using FCM algorithm, and then every subset was trained and modeled by using LS-SVM. When making forecasts, the category of the sample belongs was determined firstly. Then the corresponding sub-type model was used to predict and updated on-line. An arch dam safety warning model of deformation was established by using the proposed method. Compared with the standard LS-SVM model, the proposed hybrid model can significantly improve the generalization ability of the model.