Seismic structural health monitoring of RC framed building using artificial neural network model: a study

Presently the occurrence of strong magnitude earthquake and their effect on urban places has drawn the attention of the researchers for development of quick health monitoring technique. In response to the urgent need for quick rehabilitation of residential areas in seismic-prone regions after an earthquake, conventional methods for health monitoring of affected buildings prove time-consuming and require the assistance of competent personnel. To address this challenge, this research proposes the adoption of an Artificial Neural Network (ANN) model as a fast and robust health monitoring tool for seismic damage identification in Reinforced Concrete (RC) buildings. The developed ANN model, based on a multilayer feed-forward neural network with one hidden layer, utilizes crucial factors such as real-time earthquake ground motion (PGA, PGV, PGD, Time Duration), plinth area, and building height as inputs. Key health monitoring parameters (Inter Story Drift, Displacement, Frequency) for seismic safety are considered as outputs. Trained using the Levenberg-Marquardt algorithm and validated against new earthquake ground motion data, the ANN model demonstrates efficiency and applicability. Numerical simulations, conducted through the finite element software (ETABS) on a typical RC structure with varying plinth areas and building heights, provide the necessary data for ANN model development. Implemented in MATLAB, the results show promising potential for rapid and efficient seismic structural health monitoring, offering an effective means for post-earthquake assessment and rehabilitation of RC structures in seismically active regions.