ABSTRACT
With the monitoring data, some research work on correlation between the displacement response and wind speed has been conducted (Nakamura 2000; Xu and Chan 2009; Wang and Ding 2014). However, the formulated relations only involve the wind speed but ignore another important factor, the wind direction. From these models, the same displacement response would be produced under identical wind speeds with different wind directions. It is obviously unreasonable. In addition, these relation models were formulated by estimating the deterministic model parameters based on statistical analysis of a large amount of measured data. Different from the classical estimation method, the Bayesian approach treats the unknown para meters as distributions rather than deterministic values and more importantly, it provides room for incorporating prior information with the obtained SHM data through the Bayes theorem to achieve a feasible estimation (Zhu and Frangopol 2013). Although the Bayesian approach has been widely used in different engineering fields (Sohn and Law 1997; Enright and Frangopol 1999; Vanik 2000; Ching and Leu 2009; Yuen 2010; Au 2011; Au et al. 2012), few studies
1 INTRODUCTION
The displacement response of a long-span bridge caused by aerostatic and fluctuating wind actions, is one of important parameters characterizing the safety of the bridge under strong wind. The lateral deflection of cable-supported bridges susceptible to wind can reach a plateau. This large-amplitude displacement would threaten the safety of the whole bridge (Wang and Ding 2014). As an example, the first Tacoma Narrows Bridge in Washington State collapsed under winds of approximately 18 m/s due to violent torsional oscillations (Green and Unruh 2004).
