基于LASSO-ASAPSO-LSTM的双曲拱坝缺失位移数据恢复  

Recovery of Missing Displacement Data for Hyperbolic Arch Dams Based on LASSO-ASAPSO-LSTM

作　　者：黄民水 邓志航[1,2] 张健蔚 HUANG Min-shui;DENG Zhi-hang;ZHANG Jian-wei(School of Civil Engineering and Architecture,Wuhan Institute of Technology,Wuhan 430074,China;Hubei Provincial Engineering Research Center for Green Civil Engineering Materials and Structures,Wuhan Institute of Technology,Wuhan 430074,China)

机构地区：[1]武汉工程大学土木工程与建筑学院,湖北武汉430074 [2]武汉工程大学绿色土木工程材料与结构湖北省工程研究中心,湖北武汉430074

出　　处：《水电能源科学》2024年第12期128-132,共5页Water Resources and Power

基　　金：国家大坝安全工程技术研究中心开放基金项目(CX2023B05);水资源工程与调度全国重点实验室开放研究基金项目(2023SGG02)。

摘　　要：由于设备故障或无线传输过程中的数据包丢失等原因,存在数据缺失现象,导致大坝的安全评估无法得到保障。为此,提出了一种基于深度学习的双曲拱坝缺失位移数据恢复模型,采用最小绝对值收缩和选择算子法(LASSO回归算法)从建立的18个大坝位移影响因子中筛选出影响较为显著的环境因子;基于长短期记忆神经网络(LSTM)搭建了大坝缺失数据恢复模型;采用自适应模拟退火粒子群算法(ASAPSO)对LSTM的3个超参数进行了优化;最后,依托湖南省资兴市东江大坝累计14年(2000~2014年)的监测数据,对所提方法的计算精度和计算效率进行了验证。结果表明,ASAPSO的引入使该模型的恢复精度和效率优于常规的机器学习算法,为大坝安全监测缺失数据的准确恢复提供了有力工具。The data loss occurs due to monitoring equipment or data wireless transmission issues,which compromises the assurance of safety assessment for dams.Thus,this paper proposes a deep learning-based model for recovering miss-ing displacement data in hyperbolic arch dams.Firstly,the LASSO regression algorithm was employed to select signifi-cant environmental factors from the 18 displacement influencing factors established.Secondly,a dam missing data recov-ery model was constructed based on the Long Short-Term Memory(LSTM)neural network.Next,the adaptive simula-ted annealing particle swarm optimization(ASAPSO)algorithm was utilized to optimize three hyperparameters of LSTM.Finally,the computational accuracy and efficiency of this method were validated using 14 years(2000-2014)of monitoring data from the Dongjiang Dam in Zixing City,Hunan Province.The results demonstrate that the introduction of ASAPSO improves the recovery accuracy and efficiency of the model compared to conventional machine learning algo-rithms,providing a powerful tool for accurately recovering missing data in dam safety monitoring.

关 键 词：混凝土双曲拱坝 缺失位移恢复 长短期记忆神经网络 结构健康监测 LASSO回归 自适应模拟退火粒子群算法 

分 类 号：TV698.11[水利工程—水利水电工程]