Back analysis of construction of large deep excavations using intelligent optimization algorithm

By introducing the artificial intelligence technology, on improved Nelder-Mead acceleration algorithm based on the neural network is proposed to obtain better optimization results. A back analysis method for construction of large deep excavations using field observations is established. 3D numerical simulation analyses using the intelligent optimization technique are performed to forecast the horizontal deformation of the wall, axial force of the supports and displacement of the adjacent energy supply pipelines at later stages based on the background of excavation with an area of 93383. The results show that the convergence of the calculation is faster using the proposed method, and the number of iterations decreases by up to 86.9% compared with that of the Nelder-Mead algorithm. The predicted results are in good agreement with the monitoring data.