In this paper, a long-term probabilistic flutter analysis of long-span bridges, with consideration of the time-variant probability density function (PDF) of annual maximum wind speed caused by climate change and the deterioration effects of dynamic properties obtained by field monitoring data, is investigated. Estimated uncertainty, which occurs in probabilistic flutter analysis, is quantified by the generalized density evolution equation (GDEE). A suspension bridge, with the center span of 1650 m, is chosen as application example. Long-term deterioration and inter-seasonal varying characteristics of modal frequencies and damping ratios are discussed. An implicit formulation among radius to maximum winds $R_{\max }$, central pressure deficit $\Delta p$, latitude $\psi$ and sea surface temperature $\mathrm{SST}$ is set up by training a two-layer feed-forward artificial neural network (ANN) with historical records, and then the full-track typhoon simulation is conducted based on Vickery’s empirical model. Lastly, long-term probabilistic flutter analysis is conducted in conjunction with three prospective climate change scenarios, RCP2.6, RCP4.5 and RCP8.5, showing that the likelihood of annual flutter failure will increase greatly mainly due to higher annual maximum wind speed in a warming climate.

本文考虑了气候变化引起的年最大风速的时变概率密度函数（PDF）和现场监测获得的动力特性的恶化效应，对大跨度桥梁进行了长期概率颤振分析。数据，正在调查中。概率颤振分析中出现的估计不确定性由广义密度演化方程 (GDEE) 量化。选择中心跨度为1650 m的悬​​索桥作为应用实例。讨论了模态频率和阻尼比的长期恶化和季节间变化特性。半径到最大风的隐式公式${\mathrm{电阻}}_{\mathrm{最大限度}}$, 中枢压力不足 $\Delta 磷$, 纬度 $\psi$ 和海面温度 $\mathrm{不锈钢}$通过使用历史记录训练两层前馈人工神经网络（ANN）来建立，然后基于 Vickery 的经验模型进行全路径台风模拟。最后，结合RCP2.6、RCP4.5和RCP8.5三个前瞻性气候变化情景进行长期概率颤振分析，表明年颤振失效的可能性将大大增加，主要是由于较高的年最大风速在变暖的气候中。