Prediction of global damage index of reinforced concrete building using artificial neural network

Abstract

Determination of structural damage index (DI) is tedious and time consuming process using the available conventional methods. Traditional multivariable regression (MVR) model had been fitted by researchers to find out global damage index (GDI) of buildings although the outcome was not up to the mark. In this paper, the GDI of 18 samples of 8-storey and 12-storey reinforced concrete (RC) buildings with nine irregular plans in each height category have been predicted using artificial neural network (ANN). The results show that maximum damage occurs at the ground storey and minimum damage occurs at the top storey for all cases considered. From the MVR model, it has been found that, inter-storey drift (IDR) and stiffness are not significant input parameters whereas, joint rotation, dissipated hysteretic energy, ductility, and peak roof displacement are the significant input parameters. Based on the numerical results derived, an ANN based GDI prediction model has been derived. In comparison to MVR model, the performance of ANN model is found to be satisfactory.

Keywords:

• Artificial neural network
• local damage index
• global damage index
• multi-variable regression
• nonlinear time history analysis
• Park–Ang damage index