Earthquake Damage Classification Method based on Support Vector Machines

Petros-Fotios Alvanitopoulos, John M. Konstantinides, Ioannis Andreadis, and Anaxagoras Elenas


Seismic parameters, earthquake damage classification, damage index


This paper presents an efficient method for automatic seismic damage assessment on multistory constructions. Support Vector Machines (SVMs) are trained in order to acquire compact and efficient damage assessment models for buildings of interest. The learned models lead to drastic computational reductions of seismic damage assessment, avoiding the need for explicit calculation of the dynamic response of buildings. Comparisons with a previous study testify the effectiveness of the proposed method. The trained models are able to discriminate negligible from severe damages with 100% accuracy. Further clustering, with more damage categories, provides promising results.

Important Links:

Go Back