ABSTRACT
ANNs have become a useful method for predicting tunnel settlement. This paper presents the concept of and procedure for the Artificial Neural Networks (ANNs) method in predicting urban shield tunnel settlement. In the process, apart from soil types, buried depth, ground water table and tunnel diameter, there are many other factors have been taken into consideration, such as grouting pressure, thrust force, volume loss etc. Many existing papers have chosen to use the Finite Element Method (FEM) results as input data sets instead of real monitoring data. In this paper, the application process is based on real monitoring data obtained from a shield tunnel project in Shanghai. The results of the predicted settlement are of high quality which demonstrates its potential to be recommended as a tunnel prediction tool for similar projects in future.
