Application of data mining to lagging deformation prediction of the underwater shield tunnel

Abstract

Owing to the limitation of inner space of the shield machine, it is difficult to arrange the monitoring elements within a certain distance behind the excavation face. As a result, the deformation of the segments may not be detected in time. For underwater shield tunnels, the consequences of the lack of deformation data of the supporting structure are more serious. In this paper, the FLAC3D and the regression analysis method are used to obtain the lagging deformation data that cannot be monitored. Eight influencing factors of segments deformation and the two deformation indexes are selected to establish a BP neural network model for predicting the lagging deformation of the underwater shield tunnel. The feasibility of the prediction method is verified through engineering application. Among the three prediction methods used, namely regression analysis, neural network model, and numerical simulation, the neural network model has the highest prediction accuracy and the fastest calculation speed. Totally, the findings of this study can act as a reference for deformation prediction in similar underwater shield tunneling engineerings.