Research And Analysis About The Deformation Forecasting Methods Of Yellow River Xiaolangdi Water Hydropower Dam

According to the deformation monitoring data of Xiaolangdi water hydropower dam, this paper builds a dam deformation forecasting model by applying support vector machine (SVM). When the monitoring data is sufficient, the independent variable of SVM is restructured by using principal component analysis (PCA). When the data is lost, probability principal component analysis (PPCA) is used to recognition, in order to improve the fitting and forecast precision. Meanwhile, correlative coefficient is introduced into the selection of optimal influenced factor about the dam displacement, determining the correlative coefficients of influence factor, and the important information of different factors is obtained. The results show that the combination of the Pearson correlative coefficient and the SVM can obtain better effect. And by gray relational analysis, I select the main factor of dam deformation influence factors, which is used as input vector of SVM, making the simulation forecasting. Then the result is better than usual method.Finally, based on factors that influence GM(1,1) and SVM, the GM(1,1) model is improved and the input data of SVM model is processed. Meanwhile, the two models are combined by using weights to form GM(1,1) model of time sequence variation regression. According to the actual measured data, the GM-SVM model built by time sequence variation regression has higher accuracy of simulation and prediction.