A novel Bayesian extreme value distribution model of vehicle loads incorporating de-correlated tail fitting: Theory and application to the Nanjing 3rd Yangtze River Bridge

Vehicle loads play an important role in fatigue deterioration and overload collapses of bridges. In this paper, a novel de-correlated tail-based extreme value (EV) distribution model of vehicle load is proposed. The monitored data show that occurrences of vehicle loads are correlated. Additionally, it is more reasonable to employ the tail region of a distribution when estimating extreme loads. Moreover, a Bayesian form of this new model is constructed, and an extension of this model, the Confidence Index (CI), is defined and may be promising for applications. The monitored vehicle weight on the Nanjing 3rd Yangtze River Bridge is used to demonstrate that the proposed tail-based de-correlated EV model predicts the extreme load more accurately than traditional methods and that the Bayesian approach can further increase the precision of this estimate. Finally, the calculated CI of the complete prediction process offers a comprehensive guideline for the estimate precision.