Abstract: The soil disturbance induced by a shield tunnel can affect the bearing capacity of the existing tunnel, and the resulting vertical and horizontal deformation of the existing tunnel poses risks to normal operation. To accurately predict and effectively control the deformation of the existing tunnel induced by the undercrossing of the shield tunnel, a multiobjective optimization framework is constructed that integrates particle swarm optimization(PSO), random forest(RF), and nondominated sorting genetic algorithm Ⅱ(NSGAⅡ). A monitored sample data set is used in conjunction with the PSORF model to predict the displacement at the tunnel bottom, whereby a nonlinear mapping relationship between the main construction parameters of the shield tunnel and the bottom displacement is obtained. The constructed PSORFNSGAⅡ multiobjective optimization framework is used to optimize the bottom settlement and horizontal displacement of the existing tunnel, yielding the Pareto front of the shield tunneling parameters. The TOPSIS method is used to calculate the optimal solution to accurately predict the deformation of the existing tunnel. Optimization and control using PSORFNSGAⅡ were found to reduce the bottom settlement and horizontal displacement of the existing tunnel by 33.6% and 35.1%, respectively, thus realizing accurate prediction and control of the deformation of the existing tunnel induced by undercrossing of a shield tunnel.

Key words: shield undercrossing, existing tunnel, deformation; multiobjective optimization, prediction, control, particle swarm optimizationrandom forestnondominated sorting genetic algorithm Ⅱ