桥梁车辆荷载效应外推组合极值方法适用性比较研究

Comparative Study on Applicability of Combined Extremum Methods for Extrapolating Bridge Vehicle Load Effect

摘要:

桥梁车辆效应包括加载长度较长的整体效应和较短的局部效应，效应极值可能受到多个加载事件的影响，目前多采用组合广义极值法和组合广义帕累托分布法推定其效应极值。组合极值方法能够很好地解决传统极值理论不适用于非同一底分布样本的问题。为了进一步明确两种组合极值方法在各种车流场景中的适用范围，结合不同交通状态下的模拟车流以及某大跨径悬索桥中的主缆轴力和跨中吊杆轴力两种效应，在分析两种方法假定和算法的基础上，计算了不同加载长度下的效应极值推定结果，并对其影响因素、适用范围以及规律进行了讨论。结果表明：组合广义极值方法对样本量变化与底分布拟合较敏感，在样本量有限的情况下采用组合广义帕累托分布法更合适，在样本量充分的情况下采用组合广义极值方法计算更简便。当样本量充足时，长加载长度效应极值推定中各事件单独作用下的分布拟合曲线不存在交叉区域，其在多事件混合情况下的组合外推极值直接由控制性事件（高峰状态）产生的效应决定。

Abstract:

The effects of long-span bridges include the overall effect of long loading length and the local effect of short loading length. Their extreme value may be affected by multiple traffic loading events. Two combined methods should be used for extreme value prediction on the mixed traffic flow. The combined method can solve the problem that the traditional extreme value theory is not suitable for samples with different bottom distributions. To further clarify the application range of the two combined extreme value methods in various traffic flow scenarios， combined with the simulated traffic flow under different traffic conditions， as well as the effects on a long-span suspension bridge， based on the analysis of the assumptions and algorithms of the two methods， the effect extreme value prediction results at different loading lengths are calculated， and the influencing factors and applicable scope are discussed. The results indicate that the combined generalized extreme value method is more sensitive to the sample size change and the bottom distribution fitting. When the samples are insuffcient， the combined generalized Pareto distribution method is enjoyable. However， when the samples are sufficient， the combined generalized extreme value method tends to be easier for calculation. The distribution fitting curve of each event in the long loading length effect extreme value presumption doesn’t have a cross area on the condition of sufficient samples， and the extrapolation extreme value in the multi-event mixed situation is directly determined by the controlling event （peak status）.