Multi-objective Optimization On Excavation-induced Tunnel Displacement

The rapid development and improvement of urban rail transport has led to many urban infrastructure projects requiring excavation work adjacent to the operational metro.The excavation and unloading effect will cause additional deformation and internal forces in the tunnel structure,and in severe cases,misalignment,cracking,water seepage and track deformation of the lining pipe pieces,threatening the structural and operational safety of the metro tunnel.Therefore,controlling the disturbance of adjacent existing tunnels from pit excavation is critical to design and construction,and it is necessary to predict soil,pit and tunnel structural response prior to excavation.There are various uncertainties in foundation pit engineering,resulting in low accuracy of calculation by the traditional foundation pit forward analysis method.Based on multi-objective optimisation theory,this paper presents a multiobjective optimisation framework and an innovative optimisation algorithm for adjacent tunnel pit excavation based on multi-objective optimisation theory,oriented to consider the impact of uncertainty factors such as soil parameter uncertainty,construction conditions and time effects on the calculation accuracy,and aims to optimise the accuracy and efficiency of the calculation,in terms of both static agent model and dynamic agent model optimisation techniques.The main research work and results include.1.Relying on the case of the foundation pit of the Bund 596 project in Shanghai,numerical methods were used to simulate the construction process and to compare and analyse the ability of the soil creep principal structure model and the empirical correction formula based on actual measurements to quantify the time effect of the foundation pit.The results show that the numerical model simulates the excavation process of the pit more reasonably,however,the difference between the calculated and monitored displacement values is large,and the uncertainty of the soil parameters and the presence of the time effect make the calculation less accurate.Compared to the creep-only intrinsic model,the numerical model combined with empirical equations is more suitable as a physical model in the inverse analysis study.2.Based on the numerical model incorporating the time-effect empirical formulation,a multi-objective particle swarm algorithm(MOPSO),parametric sensitivity analysis and static proxy models were used to build a multi-objective optimisation framework for the step-by-step excavation of the foundation pit,and the tunnel displacement response values for subsequent excavation steps were predicted and updated by incorporating the previous monitoring data.The updated time effect parameters indicate that the tunnel displacements will continue to develop after the excavation is completed.For the Pareto solution of the multi-objective optimisation,a quadratic optimization technique is proposed to improve the prediction accuracy.3.A dynamic multi-objective optimisation method(DMO-AIC)based on an adaptive additive criterion is proposed,taking into account the approximate limitations of using static agent model techniques in engineering optimisation,and the feasibility of the proposed method is verified using test functions,virtual numerical models and real engineering cases respectively.The influence of the key parameters of the DMO-AIC on the calculation results is analysed and recommendations are given in the calculation of several test functions with different characteristics.In the virtual numerical model,the influence of different types of engineering errors on the multi-objective response optimisation results is analysed.In a practical engineering case,a soil model with multi-parameter inputs is considered and the updated results of the tunnel displacement response match the optimisation results of the static proxy model approach.