论文部分内容阅读
为了建立可靠的大跨桥梁全寿命温差极值分布模型,提出采用广义帕累托分布(Generalized Pareto Distribution,GPD)对超阈值温差的统计特征进行描述,并给出了超阈值温差样本相关性的去除方法和最优阈值的确定方法。为了融合温差分布的先期经验信息和不断递增的温差监测样本,建立了考虑参数更新的贝叶斯估计方法,利用Gibbs抽样对贝叶斯后验分布进行计算,进而得到准确的基于广义帕累托分布的温差极值分布模型。最后利用九堡大桥长期监测温差数据进行了验证。研究结果表明,广义帕累托分布能够对超阈值温差样本的尾部统计特征进行准确描述,提出的考虑参数更新的温差极值分布贝叶斯估计方法能够对广义帕累托分布的参数进行可靠估计,估计的统计模型比极大似然估计计算的结果更接近真实情况。研究结果可为大跨桥梁温差特性分析提供参考。
In order to establish a credible extreme temperature distribution model for long-span bridges, a generalized Pareto distribution (GPD) is proposed to describe the statistical characteristics of the over-threshold temperature difference and the correlation between the over-threshold temperature difference samples Removal method and the method of determining the optimal threshold. In order to fuse the prior experience information of temperature distribution and the increasing temperature monitoring sample, a Bayesian estimation method considering parameter update was established, and the Bayesian posterior distribution was calculated by using Gibbs sampling to get the accurate Bayesian estimation based on generalized Pareto Distribution of extreme temperature distribution model. Finally, the long-term monitoring temperature difference data of Jiubao Bridge was verified. The results show that the generalized Pareto distribution can accurately describe the tail statistical features of the over-threshold temperature difference samples. The proposed Bayesian estimation method based on temperature difference extremum distribution with parameter updating can reliably estimate the parameters of the generalized Pareto distribution , The estimated statistical model is closer to the real situation than the result of maximum likelihood estimation. The results can provide reference for the analysis of temperature characteristics of long-span bridges.