Analysis of seismic damage using time series models and Artificial Neural Networks

O.R. De Lautour, P. Omenzetter

Research output: Chapter in Book/Report/Conference proceedingChapter (peer-reviewed)

Abstract

In the past decade research into SHM systems has received much attention. One of the emerging and promising approaches is the use of time series analysis. This study develops a method for the assessment of earthquake-induced damage in buildings utilising Autoregressive (AR) time series models and Artificial Neural Networks (ANNs). AR models were applied to the computer simulated seismic response of a linear, lumped mass model of a 3-storey building under different damage conditions. Damage was simulated by a reduction in lateral stiffness at each storey. The AR coefficients were considered to be damage sensitive features of the building's response. ANNs were trained to recognize changes in the patterns of the AR coefficients caused by damage and hence identify and quantify the level of damage at each storey.
Original languageEnglish
Title of host publicationProgress in Mechanics of Structures and Materials
Subtitle of host publicationProceedings of the 19th Australasian Conference on the Mechanics of Structures and Materials, ACMSM19
PublisherTaylor & Francis
Pages889-894
Number of pages6
ISBN (Print)9780415426923
Publication statusPublished - 30 Dec 2006

Fingerprint Dive into the research topics of 'Analysis of seismic damage using time series models and Artificial Neural Networks'. Together they form a unique fingerprint.

Cite this