Machine-learning-based approach for post event assessment of damage in a turn-of-the-century building structure

The work presented in this article describes development of a machine learning (ML)-based platform for condition assessment of building structures in the aftermath of extreme events. The methods employed in the study include support vector machines, neural networks, and Gaussian Naïve Bayes techniques for training of the structural health monitoring model. The ML platform relates the change in stiffness, and strains in various structural components of the system to the intensity and location of damage. Evaluation of the proposed method was accomplished by using it for the characterization of damage in a turn-of-the-century, six-story building with timber frames and masonry walls. The building was damaged due to differential settlement of its foundation. Application of the proposed method in the building required load testing of selected structural elements, and use of the strains acquired from the field tests as input to the ML model.