Locating structural damages by matching of damage signatures utilising artificial neural networks

Research output: Journal Publications and Reviews (RGC: 21, 22, 62)22_Publication in policy or professional journalNot applicable

View graph of relations


Original languageEnglish
Pages (from-to)22-29
Journal / PublicationHKIE Transactions Hong Kong Institution of Engineers
Issue number2
Publication statusPublished - Jun 2005


In the recent years, the developments of measurement and data storage devices and high-speed computers are so rapidly that the use of measured vibration data in structural health monitoring is feasible from the hardware point of view. For example, Tsing Ma and Kap Shui Mun bridges in Hong Kong have been comprehensively Instrumented for the purpose of vibration monitoring. However, the development of software (ie the methodology for extracting useful information from the measured raw data) has not yet been sophisticated enough for structural health monitoring to put into actual applications. This paper proposes a methodology for detecting the damage locations at an early state by matching the measured damage signature to a set of calculated damage signatures utilising artificial neural networks (ANN). Note that many model updating methods can be used to estimate the damage extents If the damage locations are identified In advance. Therefore, the most difficult task in structural health monitoring is to locate all damages. An ANN model denoted as the GRNNFA, which was particularly developed for working in noisy environment, was employed as a tool for systematically matching the patterns. The proposed methodology, which consists of the damage signature matching method and the ANN design method, were verified in this paper. The results of this study have demonstrated the superior performance of applying the GRNNFA model in handling noisy training data, and the ability of the proposed methodology in structural damage diagnosis In the present of measurement noise.

Research Area(s)

  • Artificial Neural Network, Damage Signature Matching, GRNNFA, Noisy Data Classification, Structural Damage Detection