Recovery of Missing Displacement Data for Hyperbolic Arch Dams Based on LASSO-ASAPSO-LSTM
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.
万方数据
维普
