A soft computing approach to tunnel face stability in a probabilistic framework

Abstract Tunnel face is important for shallow tunnels to avoid collapses. In this study, tunnel stability studied with soft computing techniques. A database created based on the literature which used train some broadly adopted techniques, ranging from linear regression artificial neural network. The soil dry density, cohesion, friction angle, cover depth and diameter are as input parameters. techniques state whether support stable predict pressure. It found that network outperforms other pressure predicted statistically distributed samples, failure probability obtained Monte Carlo simulations. way, of can be reliably assessed estimated fairly accurately.