| Karst brings huge challenge to tunnel construction in railway,especially the tunnel water inflow,for research of its influence on construction and environment.Researchers had set up a great number of models for tunnel water inflow's prediction.The thesis relied on GaoShan tunnel?SangZhi tunnel and GaoCun tunnel of QianZhangChang railway,compared their hydrogeologic conditions(groundwater storage ability)by means of Grey correlation analysis,then predicted tunnel water inflow by means of wavelet neural network model and GA—SVM,analysed their different outcomes and discussed the validity of two models.The main research contents and conclusions are:(1)Combining the tunnel length?maximum buried depth?osmotic coefficient?annual evaporation and annual precipitation,analyzing the relativity of five parameters to water inflow by Grey correlation analysis,finding the SangZhi tunnel got the weaker groundwater storage ability,GaoShan and GaoCun tunnel got stronger groundwater storage ability.(2)Getting two main water inflow periods of two tunnels by means of wavelet analysis,the 150 d and 250 d for SangZhi tunnel,the 130 d and 250 d for GaoShan tunnel,the 70 d and 230 d for GaoCun tunnel.Then predicting water inflow during this period,finding that the wavelet neural network model got better outcomes in GaoShan and GaoCun tunnel with average RMSE of 14.28% and 11.63%,and the GA—SVM model works better in SangZhi tunnel with average RMSE of 9.35%.The thesis analyzed effective parameters of tunnel water inflow of three karst tunnels,than predicted karst tunnel water inflow by two models,completed the validity research of different models,provided some reference for research after. |
PDF Full Text Request