A neural network based methodology for estimating bridge damage after major earthquakes

Tzu-Kang Lin, Chu Chieh Jay Lin, Kuo Chun Chang

Research output: Contribution to journalArticlepeer-review

4 Scopus citations

Abstract

A neural network based seismic bridge damage estimation system is proposed to estimate damage to the bridges in Taiwan after major earthquakes. The damage estimation system is composed of two parts: a peak ground acceleration (PGA) estimator and a bridge damage estimator. The PGA neural network estimator was first trained by the ground motion data recorded from the eight largest earthquakes in Taiwan over the past ten years. The magnitude, depth, and epicenter coordinates of the earthquake were used as the inputs in this neural network. The damage estimator was trained to learn the relationship among bridge coordinates, structure types, PGAs and damage levels to bridges based on the data collected from the 1999 Chi-Chi earthquake. Sixty-four sets of bridge data were used to train a 7-8-8-2 neural network. These two neural networks were then integrated into one system and were tested, using another twenty sets of bridge damage data. The result has demonstrated that the proposed method is able to successfully assess the damage of bridges due to major earthquakes in Taiwan.

Keywords

  • Artificial intelligence
  • Bridge damage prediction
  • Major earthquake
  • Neural network

Fingerprint Dive into the research topics of 'A neural network based methodology for estimating bridge damage after major earthquakes'. Together they form a unique fingerprint.

Cite this