PRE-EARTHQUAKE ECONOMIC LOSS ASSESSMENT OF REGULAR RC FRAME STRUCTURES BASED ON MACHINE LEARNING

Earthquakes claimed severe economic losses and casualties. To assess urban earthquake damage and support disaster mitigation planning, it is necessary to conduct rapid and accurate pre-earthquake economic loss assessments for urban building clusters. For the most popularly used regular reinforced concrete (RC) frame structures in urban building clusters, this study proposed a machine learning-based method for pre-earthquake economic loss assessment. The main contents included: Establishing a seismic economic loss database for regular RC frame structures, in which the seismic responses of structural and non-structural components were considered; Based on this database, establishing the seismic economic loss prediction models for regular RC frame structures using three different machine learning algorithms. By comparing the performances of the models, it is found that the Extreme Gradient Boosting (XGB) algorithm performed well among the three machine learning algorithms in predicting economic losses, with an R² value of 0.99.