ARTIFICIAL NEURAL NETWORK-INTEGRATED BUILDING INFORMATION MODELLING FOR ENHANCED POST-DISASTER DAMAGE DETECTION AND VISUALIZATION

Abstract

This report presents a unique approach to Structural Health Monitoring (SHM) using Artificial Neural Networks (ANNs) and Building Information Modelling (BIM), designed to provide real-time damage indices as indicators of structural health. First and foremost, this research is aimed at developing resilient infrastructure in regions susceptible to natural calamities, specifically earthquakes and floods. The approach presented herein integrates advanced sensor technology and machine learning to innovate traditional SHM methods. The study first explores the data acquisition process, where an ADXL345 sensor is employed to collect acceleration data from a structure. The raw data undergoes various stages of feature extraction including detrending, filtering, Fourier Transform, frequency extraction, and calculation of mode shape coefficients. The processed data then feeds into an ANN model, which predicts the damage index, an essential parameter in SHM. The validity and robustness of the proposed methodology were confirmed through comprehensive validation, involving a case study of a bridge. Furthermore, the SHM system's integration with Autodesk Revit software allows for intuitive visualization of damage indices on the building model, thus paving the way for informed decision-making. In summary, this research has laid down a strong and versatile foundation for real-time SHM. While the focus in this research has been restricted to buildings and bridges, the developed methodology and system are adaptable to other structures, marking a significant advancement in the future of SHM.