Prediction of Seismic Performance Level of Reinforced Concrete Building Based on Structural and Seismic Parameters Combination Using Artificial Neural Network

Abstract

This study presents an analytical strategy based on Artificial Neural Networks (ANN) for analyzing the seismic performance (SP) of low-to-mid-rise reinforced concrete (RC) buildings. The performance analysis was conducted using eleven (11) three-to-twelve story RC buildings subjected to a non-linear pushover analysis (PA) based on the buildings' existing materials, loadings, and geometrical characteristics. The chosen parameters included fourteen (14) properties, eight (8) of which were geometrical and structural, and six (6) of which were seismic. The performance analysis was conducted under the guidance of the Federal Emergency Management Agency (FEMA 356) and the National Structural Code of the Philippines (NSCP 2015). The selected parameters were based on their potential influence on the performance of RC buildings during a seismic event. As the training algorithm (TA), the ANN-based Levenberg-Marquardt algorithm was used, and the hyperbolic tangent sigmoid (HTS) function was employed as the activation function (AF). The assessment demonstrated that the model attained R values of 0.9959 and 0.9887 for seismic performance at points X and Y, respectively. The aforementioned findings demonstrate the feasibility of the proposed ANN-based technique for predicting the SP of RC structures.