基于LMD-LSTM的大坝变形预测研究  

Research on Dam Deformation Prediction Based on LMD-LSTM

作　　者：徐笑笑 李冰[1] 邹聪聪 XU Xiaoxiao;LI Bing;ZOU Congcong(School of Civil Engineering and Surveying and Mapping Engineering,Jiangxi University of Science and Technology,Ganzhou 341000,China)

机构地区：[1]江西理工大学土木与测绘工程学院,江西赣州341000

出　　处：《测绘与空间地理信息》2023年第5期173-176,共4页Geomatics & Spatial Information Technology

摘　　要：针对大坝变形监测数据呈现无规律特性和传统分解方法的不足,本文基于混合模型的思想提出了基于LMD-LSTM的大坝变形混合预测模型。首先利用局域均值分解(LMD)方法将原始数据进行分解,提取出不同频率的分量,再用长短期记忆人工神经网络(LSTM)预测模型对各个分量分别进行建模,最终将各个分量的预测值叠加重构以获得大坝变形预测值。工程实例表明,LMD-LSTM模型预测结果与实际观测值拟合较好,其精度指标MAE、MAPE和RMSE分别是1.2 mm、4.12×10^(-5)%和2 mm。相较于LSTM和EMD-LSTM模型的精度指标,该文提出的混合模型预测精度更高,为大坝变形预测提供了一种新方法。In view of the irregular characteristics of dam deformationmonitoring data and the shortcomings of traditional decomposition methods,a combined prediction model of dam deformation based on LMD-LSTM is proposed.Firstly,the local mean decomposition(LMD)method is used to decompose the original data and extract the components of different frequencies.Then,the long short term memory(LSTM)prediction model is used to model each component separately.Finally,the predicted values of each component are superimposed and reconstructed to obtain the predicted value of dam deformation.The engineering example shows that the prediction results of LMD-LSTM model fit well with the actual observation values,and the accuracy indexes of MAE,MAPE and RMSE are 1.2 mm,4.12×10^(-5)%and 2mm respectively.Compared with LSTM and EMD-LSTM,the prediction accuracy of the combined model is higher,which provides a new method for dam deformation prediction.

关 键 词：大坝变形 局域均值分解 长短期记忆人工神经网络 预测精度 

分 类 号：P25[天文地球—测绘科学与技术] TB22[天文地球—大地测量学与测量工程]