ABSTRACT
This study is intended to discuss a vibration-based damage detection of pole structures. A Bayesian damage detection firstly extracts damage-sensitive features from the autoregressive model from acceleration measurement of the pole structure, and then detects changes in the damage-sensitive feature by a Bayesian hypothesis testing. As damage scenarios in the full-scale laboratory experiment, three levels of anchor bolt tension loosening at the baseplate were considered. To clarify vibration characteristics by installed five tri-axial accelerometers, an impact test was conducted in each level of anchor bolt tension loosening. Furthermore, in order to examine practical applicability of the Bayesian damage detection method using a limited number of sensors, the feasibility of detecting damages using acceleration data from a single sensor installed at an accessible location by users is also investigated. Observations demonstrated that the proposed method can detect damages at an earlier stage before detectable changes in natural frequency of the pole structure.
