Statistical Estimation of Seasonal Variations in Measured Track Geometry

Abstract

This paper describes a method of statistically extracting seasonal variations from data sets of vertical track geometry measured at the same point for years. Track geometry changes seasonally at some fixed points although track geometry is generally considered to change monotonically because of daily train loads. The statistical time-series model for extracting a seasonal variation is developed and the parameters included in the model are estimated according to the maximum likelihood method.