Extreme Value Estimation and Research on Dynamic Coefficients of Wheel Rail Force Based on GPD Theory and Percentile Threshold Method
The design value of train load is one of the key design parameters in the design and calculation of the ballast-less track structures of high-speed railways.Based on the generalized Pareto distribution(GPD)theory,this paper stud-ied the extreme estimation methods of wheel rail force excited by whole wavelength range irregularity,and compared the results with those under pulse excitation,providing a basis for the design of ballastless track structures and improvement of relevant codes.The results show that when percentile threshold method is used to select thresholds,samples should be declustered to improve the independence between samples and the accuracy of extreme estimation of wheel rail force.By proposing a shape parameter screening method for the extreme estimation of wheel rail force combining percentile thresh-old,the corresponding sample size and shape parameter range are determined for each cluster.It is recommended that sample size is 3×105～5×105,with the recommended percentile threshold of 50％～98％.Additionally,it is necessary to use the mean of the extreme estimations as the final extreme estimation of wheel rail force.The dynamic coefficients of train load design value for train speeds ranging from 250 km/h to 400 km/h are 2.2,2.4,2.6 and 2.8,respectively.
generalized Pareto distribution theorywheel-rail forceextreme value estimationshape parameterdynamic coefficient
万方数据
维普
