ABSTRACT
Fatigue is an important safety indicator of bridge member under long-term random live loads. This paper focuses on the bridge member’s fatigue assessment caused by traffic and wind loads. The numerical model of the fatigue analysis of the bridge-traffic-wind system is implicit, which includes several time-progressive sub-modules. A new fatigue reliability assessment model is constructed based on Mind Evolutionary Algorithm (MEA). The Back Propagation (BP) neutral network optimized by MEA can be used to generate the limit state function by rationally considering all the time-progressive random variables. Particle Swarm Optimization (PSO) is then applied to improve the traditional Monte Carlo method, which can help choose samples with high efficiency. Through applying the new reliability-computation model to a typical cable-stayed bridge under variable traffic and wind loads, it is found the new method can compute the fatigue reliability index with higher efficiency compared to tradition model.
