Analysis of factors influencing tunnel deformation in loess deposits by data mining: A deformation prediction model

摘要

AbstractDue to the special properties of loess, the deformation of tunnels constructed in loess is generally large and easily induced. To control deformation during construction, the degree of influence of multiple factors on tunnel deformation is analyzed by data mining and a deformation prediction model is established, based on tunnels along the Menghua railway of China. Both objective environment and manual operation are considered. The surrounding rock level, groundwater condition, burial depth, excavation method and support close time are selected as the main factors influencing tunnel deformation. The influence degree of each factor is calculated through mining statistical data collected from the project. Finally, using influencing factors as evaluation indices, a Rough set-extension model for predicting loess tunnel deformation is established and tested. Results obtained via the prediction model are in good agreement with field observations. The study quantifies the influence degree of each selected factor on deformation of the loess tunnel, which in turn can help in deformation control efforts. Moreover, the Rough set-extension model realizes a multi-criteria prediction of the loess tunnel's deformation and provides a practical guide for construction of similar projects.Highlights?Deformation of excavated sections of the loess tunnel is classified by the Delphi-extension model based on monitoring data.?Influence degree of factors on deformation is analyzed through mining statistical data by rough set.?Rough set-extension model for predicting the loess tunnel deformation is established.]]>