Back analysis of rock mass parameters in mechanized twin tunnels based on coupled auto machine learning and multi-objective optimization algorithm
Chengwen Wang, Xiaoli Liu, Jiubao Li, Enzhi Wang, Nan Hu, Wenli Yao, Zhihui He
Abstract
Accurate determination of rock mass parameters is essential for ensuring the accuracy of numerical simulations. Displacement back-analysis is the most widely used method; however, the reliability of the current approaches remains unsatisfactory. Therefore, in this paper, a multistage rock mass parameter back-analysis method, that considers the construction process and displacement losses is proposed and implemented through the coupling of numerical simulation, auto-machine learning (AutoML), and multi-objective optimization algorithms (MOOAs). First, a parametric modeling platform for mechanized twin tunnels is developed, generating a dataset through extensive numerical simulations. Next, the AutoML method is utilized to establish a surrogate model linking rock parameters and displacements. The tunnel construction process is divided into multiple stages, transforming the rock mass parameter back-analysis into a multi-objective optimization problem, for which multi-objective optimization algorithms are introduced to obtain the rock mass parameters. The newly proposed rock mass parameter back-analysis method is validated in a mechanized twin tunnel project, and its accuracy and effectiveness are demonstrated. Compared with traditional single-stage back-analysis methods, the proposed model decreases the average absolute percentage error from 12.73% to 4.34%, significantly improving the accuracy of the back-analysis. Moreover, although the accuracy of back analysis significantly increases with the number of construction stages considered, the back analysis time is acceptable. This study provides a new method for displacement back analysis that is efficient and accurate, thereby paving the way for precise parameter determination in numerical simulations. • A mechanized twin tunnels parametric modeling method based on EPBM-hydraulic-rock-structure coupling is proposed, and a corresponding construction response simulation platform is developed. • A multi-stage rock mass parameter back analysis method based on AutoML-MOOA considering the construction process is proposed, which has good accuracy and effectiveness. • The accuracy of multi-stage back analysis method based on AutoML-MOOA is much higher than that of traditional single-stage back analysis. • The more excavation stages used, the higher the accuracy of the parameter back analysis.