BACK ANALYSIS OF LINEAR AND NONLINEAR GROUND BEHAVIORS AROUND SHALLOW TUNNEL BY USING ARTIFICIAL NEURAL NETWORK

Abstract

Artificial Neural Network is used as a key technique for back analysis in a nonlinear tunneling problem. Database is created by performing series of nonlinear finite element analysis while varying parameter values within reasonable ranges determined from experience. Neural networks created from this database are found to be able to identify parameters concerned with both linear and nonlinear deformation behaviors of shallow tunnel. The exercises conducted here show that the proposed method has reasonable stability in all aspect of data preparation and computation and could be a new engineering tool for modern tunnel engineers.