水利水电技术2018，Vol.49Issue(4)：45-49,5.DOI:10.13928/j.cnki.wrahe.2018.04.007
基于PCA-RBF神经网络的混凝土坝位移趋势性预测模型
PCA-RBF neural network-based prediction model fordisplacement trend of concrete dam
陈斯煜 1戴波 2林潮宁 3曹文翰1
作者信息
- 1. 河海大学 水利水电学院,江苏 南京 210098
- 2. 河海大学 水文水资源与水利工程科学国家重点实验室,江苏 南京 210098
- 3. 河海大学 水资源高效利用与工程安全国家工程研究中心,江苏 南京 210098
- 折叠
摘要
Abstract
In order to improve the prediction accuracy of displacement trend for concrete dam,a prediction model (PCA-RBF)based on principal component analysis (PCA) and radial basis function (RBF) neural network for analyzing the displacement trend of concrete dam is established herein.Firstly,the dimension reductions are made on the multi-observation point radial displacement monitoring data with the principal analysis,so as to eliminate the multi-correlation that influences the relevant component dataset and then the principal element displacement and the principal influence component are extracted respectively.Secondly,both the principal element displacement and principal influence components are put into the radial basis function neural network to build the prediction model for predicting the extracted principal element displacement.Finally,the result from the application of this method to a concrete dam shows that the root-mean-square error (RMSE),the mean absolute error (MAE) and the mean absolute percentage error(MAPE) of the PCA-RBF model are 2.037 8 mm,1.698 6 mm and 3.32％ respectively,which are significantly smaller than those from the conventional multiple regression model (MRM),the radial basis function (RBF) neural network model and the BP neural network model (PCA-BP),for which the factors are processed with the principal component analysis,thus it is reflected that the PCA-RBF model proposed herein has a better prediction accuracy.关键词
主成分分析/径向基神经网络/混凝土坝/位移预测模型/大坝安全监测Key words
principal component analysis/radial basis function neural network/concrete dam/displacement prediction model/dam safety monitoring分类
水利科学引用本文复制引用
陈斯煜,戴波,林潮宁,曹文翰..基于PCA-RBF神经网络的混凝土坝位移趋势性预测模型[J].水利水电技术,2018,49(4):45-49,5.
