engleski

A machine learning approach to the seismic fragility assessment of buildings

In the context of Performance-Based Earthquake Engineering, an intensity measure provides a link between the probabilistic seismic hazard analysis and the probabilistic structural response analysis. The purpose of this study is to develop a structural damage classifier and improve current prediction on the basis of a given intensity measure and different supervised machine learning algorithms: Support-Vector Machine, Logistic Regression and Random Forest. In particular, the efficiency of four different IMs for estimating the seismic response of three different kind of buildings is evaluated, namely peak ground acceleration, spectral acceleration evaluated at the principal period, average spectral acceleration and filtered incremental velocity. The classifier will be able to predict the post- earthquake damage state, given the geometry of the building and the intensity of the ground motion input. In particular, the purpose of this classifier is to accelerate post- earthquake damage evaluation of critical buildings. This will allow faster recovery time and decrease financial losses expected from downtime and repair. A focus is made on three different buildings typologies that can be used to represent the majority of the building stock in the city of Ferrara (Italy).

Performance-Based Earthquake Engineering ; Machine learning ; Seismic fragility assessment ; Buildings

nije evidentirano