The fatigue loads on the carbody of a high-speed train (HST) increase gradually with operating mileage. Because of the long design life of the carbody, it is difficult to conduct a long-term test of its fatigue load history. Instead, the loads are usually measured over relatively short periods of time and then extrapolated. This paper extracts several representative load-history samples during the service cycle of carbody of HST and base on the rainflow-counting matrix of these samples. The method of two-dimensional variable-bandwidth kernel density estimation (KDE) is used to extrapolate the HST carbody loads. Finally, the KDE extrapolation data are verified against measured HST data and compared with the results of conventional linear extrapolation. The finding is that KDE extrapolation based on multi-sample load data can give results that are similar to the measured data. Compared with traditional linear extrapolation, KDE extrapolation is better at simulating low-cycle large-scale loads. Comparing the damage calculated by KDE extrapolation with the measured data, the error between the two is 6.59%, indicating that KDE extrapolation meets the mechanical engineering requirements.
Application of kernel density estimation to extrapolating the fatigue loads on a high-speed train
Vehicle System Dynamics ; 58 , 8 ; 1212-1225
2020-08-02
14 pages
Aufsatz (Zeitschrift)
Elektronische Ressource
Unbekannt
High-speed, high-density train allocation
British Library Conference Proceedings | 1994
|Extrapolating HST Lessons to NGST
British Library Online Contents | 1998
|