采用广义Pareto分布进行车辆荷载效应极值估计的研究

Extreme estimation for vehicle load effect based on generalized Pareto distribution

为了解决车辆荷载效应数据间的相关性问题并能够充分利用样本数据,提出了一种改进的独立风暴法。首先提出了一种三次类阈值法进行数据的初步分析来获得阈值;然后考虑到简支梁桥的车辆荷载效应主要受单辆重车影响,提出了改进的独立风暴法进行独立同分布样本数据的提取;最后采用广义Pareto分布进行车辆荷载效应极值估计;利用该方法,对动态称重系统记录的国内某大桥实测数据进行了车辆荷载效应极值估计,并与超越阈值法、独立风暴法进行了对比分析,结果表明:在较短评估期(T<20年)内,三种方法均可以较好地预测荷载效应极值;而在中长评估期内,改进的独立风暴法预测值较高,预测结果偏于安全。

In order to solve data correlation problems for vehicle load effect and make full use of data samples,the modified method of independent storms（ MMIS） was proposed. Firstly,a triple-class threshold method was proposed to obtain the threshold with the primary data analysis. Then,considering that the vehicle load effect of simply supported beam bridges was mainly affected by a single heavy vehicle,the MMIS was adopted to extract the independent and identically distributed（ IID） sample data. Finally,the extreme vehicle load effect was estimated with the generalized Pareto distribution. In the end,the estimation of the extreme vehicle load effect was performed for the measured data of a bridge recorded with the weighinmotion（ WIM）. The results were compared with those of the peak-over-threshold method and those of the method of independent storms. The results showed that within the shorter estimation period（ T 20years）,all the three methods can be used to better predict the extreme load effect; while within the middle or longer estimation period,the estimation of the modified method of independent storms is higher and safer.