Prediction of Nonlinear Drift Demands for Buildings with Recurrent Neural Networks

14th INTERNATIONAL CONGRESS ON ADVANCES IN CIVIL ENGINEERING, 6 - 08 Eylül 2021, cilt.1, ss.1287-1294

Application of deep learning algorithms to the problems of structural engineering is an emerging research field. In

this study, a deep learning algorithm, namely recurrent neural network (RNN), is applied to tackle a problem

related to the assessment of reinforced concrete buildings. Inter-storey drift ratio profile of a structure is a quite

important parameter while conducting assessment procedures. In general, procedures require a series of time

consuming nonlinear dynamic analysis. In this study, an extensive RNN is trained to tackle these problems and

provide a simple tool for assessment. Aim of the study is to predict the non-linear drift demand along the height

of a structure by employing RNN for a given stiffness profile along the height, strength reduction coefficient, mass

density on a floor, number of storeys. In order to train the network, a large number of nonlinear time history

analyses are conducted for synthetically created building models. It is shown that RNN is able to accurately predict

nonlinear drift demand profile of a structure along height without conducting tedious time history analyses.

Therefore, the trained RNN can serve as a drift demand estimation tool, significantly shortening the assessment

procedure.