Neural network damage detection in a bridge element

Smart structures technology is being increasingly applied to civil structure applications. In particular, development of health monitoring for bridge structures is of considerable importance. In order to explore the possibility of developing such a system, an investigation was carried out on a scale model steel bridge element using an attached sensor system consisting of two point sensors (piezoelectric accelerometers) and one integrating sensor (fiber optic modal sensor). The model element was selectively configured to produce the equivalent of a number of damage conditions. For each condition, it was physically perturbed. The sensor outputs were then used as inputs to a neural net which then provided an estimate of structural damage. A reasonable correlation between net output and actual damage indicated that this type of health monitoring system offers potential for practical application on full scale bridge structures.