Author(s)

H. Hagiwara & A. Mita

Abstract

Local damage assessment of a building using Support Vector Machine H. Hagiwara & A. Mita Department of System Design Engineering, Keio University Abstract A damage detection method utilizing the Support Vector Machine (SVM) is proposed. The SVM is a powerful pattern recognition tool applicable to complicated classification problems. Modal frequencies of a structure are used for pattern recognition in the proposed method. Typically, only two vibration sensors detecting a single input and a single output for a structural system can easily determine modal frequencies. For training SVMs the relationship between changes normalised by original modal frequencies, before suffering any damage, is utilized. The SVM trained for single damage was also found to be effective for detecting damage in multiple stories. The SVM based damage assessment is able to identify damage qualitatively as well as quantitatively. 1 Introduction Many approaches for structural health monitoring (SHM) have been proposed for

Keywords