Artificial Intelligence-Assisted Optimization of Tunnel Support Systems Based on the Multiple Three-Dimensional Finite Element Analyses Considering the Excavation Stages

The increased processing capacity of computer systems has enabled the realistic analysis of complex geotechnical problems with the finite element method and detailed optimization studies. In this paper, the Boztepe Tunnel was chosen as an example to determine the optimization solutions of the general tunnel structures, and numerous three-dimensional finite element analyses were conducted to optimize the engineering properties of the tunnel support systems. Optimization analyses were performed using the response surface method and the novel artificial intelligence-assisted improved goal attainment method. For the first time in the literature, two different optimization methods were compared with each other, and the most appropriate solution system with the highest desirability value was designated. The elasticity modulus of the rock bolts was reduced by 19% for the tunnel’s T1-Pn sector and 15% for the T1-4 sector in the preferred solution system without exceeding the maximum deformation value according to the current condition. In addition, the number of rock bolts for the T1-Pn and T1-4 sectors was decreased by 7.7% and 11.8%, respectively. As a result of the optimization studies carried out depending on the multiple three-dimensional finite element analyses conducted, safe and economical design parameters were achieved for the general tunnel models utilizing the support system.