Enhancing seismic vulnerability assessment: a neural network effort for efficient prediction of multi-storey reinforced concrete building displacement

The present paper explores the use of Artificial Neural Networks (ANN) to predict storey displacement in multi-storey reinforced concrete buildings by analysing structural parameters that significantly influence storey displacement. The present adopted technique for non-linear dynamic analysis is time- and resource-consuming and requires high technical skills for accurate results for multi-storey buildings. Therefore, the present study focuses on establishing correlations between structural parameters and storey displacement to enable fast-track estimation of seismic responses. Modal response spectrum analysis is used to simulate earthquake loading and produce storey displacement data for training the neural networks. Furthermore, varying data sizes of training and testing data are utilized for developing high-performing machine learning models and the best-performing model is identified. The results indicate that ANN with 60% data in training and 40% in testing has the best accuracy of 99% in both training and testing phases. The study's findings can aid in earthquake vulnerability assessment, understanding building conditions due to earthquake loads, and recommending building maintenance accordingly. Additionally, the results can act as a preliminary analysis for earthquake loadings allowing designers to proceed with the detailed structural design based on acceptable storey displacement results. © The Author(s), under exclusive licence to Springer Nature Switzerland AG 2023.