Fuzzy pattern recognition in vibration-based structural health monitoring

Abstract

In this paper, fuzzy sets are established for different states of a reinforced concrete bridge based on finite element model of the bridge. This bridge has been monitored continuously using vibrational signals obtained from accelerometer and strain gauge sensors installed on girders of the bridge. A Finite element model of the bridge is calibrated based on real data gathered from the bridge to be a close representation of the real structure. The calibrated finite element models of the bridge are then constructed for healthy, medium damaged and severe damaged states of the bridge. Using fuzzy set principles, the healthy, medium and severe damage states of the bridge are constructed with fuzzy bell functions. Fuzzy pattern recognition using similarity between fuzzy sets is then utilized to identify any unknown states of the bridge that can give authorities an unbiased tool for efficient maintenance of the bridge.