Application of Artificial Neural Networks to Identify Earthquake-Induced Structural Damage in Residential Buildings

Abstract

Structural damage in residential buildings can significantly compromise the safety of occupants, reduce the structural integrity of the building, and shorten its overall lifespan, potentially leading to costly repairs, decreased property value, and increased risk during natural disasters such as earthquakes or storms. Traditional methods for assessing structural damage primarily depend on manual inspections, which are not only time-consuming and labor-intensive but also susceptible to human error, subjectivity, and inconsistency. These limitations can lead to delayed detection of critical issues, inaccurate assessments, and increased risk of overlooking hidden or early-stage damage, ultimately compromising the effectiveness of maintenance and safety measures. By employing a comprehensive dataset that includes a range of structural characteristics and damage indicators, this study trains a neural network model to identify and learn patterns linked to structural damage. This study investigates earthquake-induced damage in different structural components of residential buildings, employing hundreds of feature sets including building height, number of floors, earthquake intensity, damping ratio, crack location, and material properties to train and validate a network for damage prediction. The performance of the ANN is evaluated, demonstrating superior accuracy and efficiency. The results highlight the potential of ANNs to revolutionize structural health monitoring by providing rapid, cost-effective, and reliable damage assessments, thereby enhancing preventative maintenance and mitigating risks associated with structural failures.