Predictive railway maintenance based on statistical analysis of track geometric parameters

The current trends in transport require more competitive and self-financing transport modes hence railway companies must fulfil the criteria by lowering their expenditure and still meet the customers' requirements. This paper proposes a methodology for track maintenance with the aim of being able to perform maintenance works only when they are strictly necessary while keeping safety and comfort levels. Such a methodology is based on a predictive maintenance philosophy and on a statistical approach. The characterisation of the track defects by means of probability functions, the obtaining of the most appropriate measuring step for data study and the temporal analysis of consecutive track surveys are discussed in the present research. Finally the paper provides some promising results. © Civil-Comp Press, 2012