Seite - 913 - in Book of Full Papers - Symposium Hydro Engineering

Table 3 Evaluation of the related parameters Stage Model R2 RMSE/mm MAE/mm Training PCA-RBF 0.959 0.288 0.201 Statistical 0.958 0.293 0.209 Prediction PCA-RBF 0.910 0.702 0.548 Statistical 0.869 0.827 0.696 4. CONCLUSION In this paper, a monitoring model based on the principal component analysis (PCA) and radial basis function (RBF) neural network is put forward to predict the displacement trend of the concrete dam. And an example analysis based on the proposed model is performed on a prototype concrete dam. The conclusions are as follows: (1) The PCA-RBF prediction model reduces the data redundancy and eliminates the multiple correlations among the components. And this model can predict the displacement trend of the concrete dam well. (2) Compared with the statistical model, the PCA-RBF neural network model has higher prediction accuracy for the displacement trend of concrete dam, and can be applied in practical projects. ACKNOWLEGMENTS This research has been greatly supported by the National Key Research and Development Plan (Grant No. 2016YFC0401601), Postgraduate Research & Practice Innovation Program of Jiangsu Province（KYCX17_0436）and the Fundmental Research Funds for the Cental Universities (2017B623X14). 913