Estimation of lateral wall movements in braced excavations using neural networks

The growing interest in neural networks among geotechnical engineers is due to its excellent performance in modelling nonlinear multivariate problems. This paper demonstrates that neural networks can synthesize data derived from finite element studies on braced excavations in clays and capture the nonlinear interactions between the variables in the system. The neural network was able to produce reasonably accurate wall displacement predictions after "learning" from examples derived from finite element analyses. This method has the advantage over other more conventional methods in that it can be readily retrained as additional data from finite element studies and actual field records are acquired.