中文摘要：

为了根据实测车流数据准确评估钢桥面板的疲劳安全水平,提出了基于随机车流荷载的疲劳可靠度评估框架。建立了考虑车流参数的疲劳损伤功能函数和疲劳应力概率模型,应用均匀设计-支持向量回归（UD-SVR）方法解决随机车流在桥梁有限元应力分析中的耗时问题,分析了交通量和轴重增长系数对钢桥面板焊接细节疲劳可靠度的影响规律。结果表明：基于动态称重数据的随机车流模型在钢桥疲劳应力概率建模方面具有一定的适用性;基于UD-SVR的有限元模型替代方法能够提高疲劳应力概率建模的效率;行车道处桥面板的日等效应力幅远大于超车道,但应力循环次数略小于超车道;车辆轴重增长系数对疲劳可靠指标的影响远大于交通量增长系数的影响。

英文摘要：

In order to accurately assess fatigue safety level of steel bridge decks according to the measured traffic flow data,an assessment framework on fatigue reliability based on stochastic traffic flow load was proposed.The fatigue damage function and probabilistic model of fatigue stress were established with consideration of traffic flow parameters.A uniform design-support vector regression（UD-SVR）method was applied to solve the time-consuming problem in finite element analysis of bridge stress under random traffic flow.Influence laws of growth coefficients of both the traffic volume and axle weight on fatigue reliability of the welded details of the bridge deck were analyzed.The results show that the weigh-in-motion（WIM）-based random traffic flow model is applicable for probability modeling of fatigue stresses of steel bridge to some extent;the substitute method of finite element model based on UD-SVR method can improve the efficiency of probabilistic modeling for fatigue stress.The daily equivalent stress range in bridge deck of traffic lane is greater than that of the fast lane,but the corresponding number of cycles is slightly less than that of the fast lane.The growth coefficient of axle weight has a greater influence on fatiguereliability index,compared with that of the traffic volume.