基于EEMD-LSTM的桥梁变形响应组合预测模型研究  

Combined Prediction Model of Bridge Deformation Response Based on EEMD-LSTM

作　　者：孟庆成[1] 李明健 胡垒 万达 吴浩杰 齐欣[2] MENG Qingcheng;LI Mingjian;HU Lei;WAN Da;WU Haojie;QI Xin(School of Civil Engineering and Geomatics,Southwest Petroleum University,Chengdu 610500,Sichuan,China;School of Civil Engineering,Southwest Jiaotong University,Chengdu 610031,Sichuan,China)

机构地区：[1]西南石油大学土木工程与测绘学院,四川成都610500 [2]西南交通大学土木工程学院,四川成都610031

出　　处：《重庆交通大学学报（自然科学版）》2023年第5期25-34,共10页Journal of Chongqing Jiaotong University(Natural Science)

基　　金：国家自然科学基金项目(52078442,51408498);四川省科技计划项目(2021YJ0038);四川省教育厅自然科学重点项目(16ZA0058)。

摘　　要：为准确预测桥梁结构变形响应,提出了一种利用孤立森林(iForest)算法对桥梁原始变形数据进行降噪,集合经验模态分解法(EEMD)对桥梁变形数据进行分解,长短期记忆神经网络(LSTM)深度学习法对所得到的多尺度变形分量进行预测的EEMD-LSTM组合模型。以武汉沌口长江大桥作为研究对象,选取RMSE、MAE、MAPE和R^(2)等参数作为评价指标,对该模型进行了验证。研究结果表明:与单一的LSTM、SVM和Bayesian模型相比,EEMD-LSTM模型有着良好的鲁棒性、适用性和更高的预测精度。In order to accurately predict the deformation response of bridge structure,an EEMD-LSTM combined model was proposed,in which the iForest algorithm was used to denoise the original deformation data of the bridge,the Ensemble Empirical Mode Decomposition(EEMD)method was used to decompose the bridge deformation data,and the Long Short-Term Memory(LSTM)deep learning method was used to predict the obtained multi-scale deformation components.Taking Wuhan Zhuankou Yangtze River Bridge as the research object,RMSE,MAE,MAPE and R^(2)were selected as evaluation indexes to verify the proposed model.The research results show that compared with the single LSTM,SVM and Bayesian model,the EEMD-LSTM model has good robustness,applicability,and higher prediction accuracy.

关 键 词：桥梁工程 深度学习 孤立森林算法 EEMD-LSTM 变形响应 

分 类 号：U446[建筑科学—桥梁与隧道工程]