HEALTH STATUS PREDICTION SYSTEM FOR STRUCTURAL HERITAGES AND BUILDINGS: A COMPARATIVE ANALYSIS USING A DEEP AND MACHINE LEARNING MODEL

ABSTRACT

Early detection of structural damage is essential for the maintenance, repair, and rehabilitation of the building. Yet, data science has emerged as a significant and useful technique for a number of civil engineering applications, such as structural health monitoring (SHM). As a result, the goal of this effort was to use a machine and deep learning model to forecast a building’s real-time health state. Support Vector Machine (SVM), and Convolutional Neural Network are the models that were used to create the building health prediction status system. Four classifications of building health were made: crack, efflorescence, intact, and spalling. The dataset that was utilized to train the models contains 21,092 images. The accuracy for CNN and SVM were compared and gave a value of 71% and 50% respectively thus showing that the deep learning technique performed better than the machine learning model when these models were tested.

Keywords: Machine Learning, Deep Learning, CNN, SVM, Building Health, Structural Health Monitoring, Damage Detection