Construction Parameter Optimization for Shield Tunneling Based on the Multi-Level ANN

Setting appropriate construction parameters to control the ground settlement is the linchpin of shield tunneling. This paper divides the whole shield drive process into seven stages, in which each stage is simulated by an artificial neural network. The multi-level artificial neural network, which is composed by the seven stages, builds the relational model between construction parameters and the ground settlement. According to this model, we optimize the control scheme to ensure quality, increase speed and decrease cost by genetic algorithms. This approach has been applied to many tunneling projects and the results show good prospect.