在役桥梁动应变长期监测数据的极值估计

发布时间：2018-02-25 23:28

本文关键词： 桥梁健康监测应变极值滤过的泊松过程模拟退火算法概率模型广义Pareto分布　出处：《铁道学报》2017年11期 　论文类型：期刊论文

【摘要】：车辆荷载效应是桥梁安全评估及可靠性研究的关键因素,基于长期应变监测数据的应变极值估计很有意义。以往估计车辆荷载效应模型的观测样本相对较少,常用的统计模型对于样本高分位点的估计也不够准确。采用太平湖大桥120d的动应变监测数据,选取车辆荷载作用下应变峰值为研究样本,确定所需最小采样时长。采用广义的Pareto分布作为样本超阈值概率模型,滤过的泊松过程作为样本超阈值随机过程的概率模型,结合极值统计理论,估计桥梁剩余服役期内车辆荷载引起的应变极值,给出桥梁不同承载能力情况下主梁的可靠指标及失效概率。结果表明:广义的Pareto分布模型可较好地拟合出桥梁动应变的右尾分布,能够比较准确地得出车辆荷载作用下应变极值的概率分布。
[Abstract]:Vehicle load effect is a key factor in bridge safety assessment and reliability research. It is significant to estimate the maximum strain value based on long-term strain monitoring data. The commonly used statistical model is also not accurate for the estimation of high score sites of samples. The dynamic strain monitoring data of Taipinghu Bridge for 120 days are used, and the peak strain under vehicle load is selected as the research sample. Using the generalized Pareto distribution as the sample super-threshold probability model and the filtered Poisson process as the probabilistic model of the sample super-threshold stochastic process, combining with the extreme value statistics theory, the minimum sampling time is determined. The strain extremum caused by vehicle load during the remaining service period of the bridge is estimated. The reliability index and failure probability of the main beam under different load-bearing capacity are given. The results show that the generalized Pareto distribution model can fit the right-tail distribution of the bridge dynamic strain well. The probability distribution of strain extremum under vehicle load can be obtained more accurately.
【作者单位】：合肥工业大学土木与水利学院;中南大学土木工程学院;
【基金】：中国博士后科学基金(2015M581982) 安徽省自然科学基金(E080505)
【分类号】：U441;U446

【相似文献】