Development of a damage evaluation tool for non-instrumented buildings using machine learning

Visual building damage inspections conducted after strong earthquakes can take weeks to months to be completed. While methods that use floor acceleration data to estimate damage are available, only a small portion of buildings are instrumented. To achieve rapid damage assessments over a wide area, methods to estimate damage to non-instrumented buildings are required. This study details ongoing research to estimate the response of non-instrumented buildings as an intermediary step to develop a regional-level building damage assessment.