Abstract
The Jigongshan Mountain Tunnel in Shenzhen China is 4.6 km in length and designed to cross underneath the Xiaping Municipal Solid Waste Landfill. The estimation of the in situ stress field along the tunnel, particularly under the Xiaping landfill, has posed significant challenges. In this study, an integrated strategy comprising multisource information analysis and intelligent back analysis was developed to predict the in situ stress of the study region. First, the regime of in situ stress was estimated based on multisource information, including geological history, geological structure, topography, rock type, joint, and measured stress. Then, multiple linear regression and particle swarm optimization (PSO) method coupled with extreme learning machine (ELM) method were introduced into the intelligent back analysis to identify the orientation and magnitude of the in situ stress along the tunnel. By simultaneously considering the effects of gravity, tectonic stress, rock and fault parameters, and landfill load on the in situ stress, the sum of squares for error in simulation of in situ stress significantly decreases by 23%. Finally, the accuracy of the predicted in situ stress field and the reliability of the proposed integrated method were substantiated through a comprehensive comparison between the simulated and measured rock displacement acquired during the tunnel excavation process. The characteristics of the in situ stress along the tunnel and under the landfill were successfully identified. The landfill load predominantly influences the vertical stress under the landfill, while it does not alter the three-dimensional state of stress. This study serves as a valuable reference for identifying the in situ stress in regions with varying upper loads and limitations on conducting field measurements.