Dynamic soil–structure interaction analysis of buildings by neural networks

In this study, neural networks (NN), which have been widely applied to engineering problems, have been applied to soil–structure interaction (SSI) problems. In the modelling of the NN, the back propagation algorithm has been used. The training and testing sets have been developed by using finite element (SAP2000) package program outcomes. Various analyses have been performed for different height/width ratios of buildings in terms of different soil properties, local site depths, and different shear wave velocities. These effects to the periods, top stories acceleration and displacement have been presented. It is observed that the modelling and analysis stages are minimized, the solution time is quite fast and the NN has shown a satisfactory performance for the solution of SSI problems. The NN results are compared with SAP2000 outcomes and found to be encouraging.