排灌机械工程学报2026，Vol.44Issue(3)：292-299,8.DOI:10.3969/j.issn.1674-8530.24.0043
基于DenseNet-GRU的混凝土拱坝变形深度学习预测模型
Deep learning prediction model for concrete arch dam deformation prediction based on DenseNet-GRU
摘要
Abstract
Traditional prediction methods based on statistics or machine learning often struggle to effec-tively capture the complex mapping relationships between the displacement of concrete arch dams and various influencing factors.Thus,a novel deep learning prediction method was proposed.This method integrated densely connected convolutional networks(DenseNet)and gated recurrent unit(GRU)to form a DenseNet-GRU model,aiming to enhance the accuracy and generalization ability of deformation prediction for concrete arch dams.A typical concrete arch dam located in a certain region of China was selected as a case study,and deformation monitoring data from multiple measuring points were used for empirical analysis.The results indicate that the DenseNet-GRU model can accurately simulate the dis-placement deformation process of all monitoring points.Compared with other deep learning models,it demonstrates higher prediction accuracy and stronger generalization capabilities.This research provides an efficient and reliable prediction tool for dam safety monitoring and health management,and holds sig-nificant theoretical and practical implications for the advancement of dam safety management practices.关键词
混凝土拱坝/大坝变形预测/深度学习模型/密集连接卷积网络/门控循环单元神经网络Key words
concrete arch dams/dam deformation prediction/deep learning model/densely connected convolutional network/gated recurrent unit neural network分类
农业科技引用本文复制引用
刘宇星,柴军瑞..基于DenseNet-GRU的混凝土拱坝变形深度学习预测模型[J].排灌机械工程学报,2026,44(3):292-299,8.基金项目
国家自然科学基金资助项目(51679197) （51679197）
陕西省创新团队项目(2022TD-01) （2022TD-01）
