Hamid Khajehei - Division of Operation and Maintenance, Luleå University of Technology, Luleå, Sweden
Alireza Ahmadi - Division of Operation and Maintenance, Luleå University of Technology, Luleå, Sweden
Iman Soleimanmeigouni - Division of Operation and Maintenance, Luleå University of Technology, Luleå, Sweden

Accurate prediction of the track geometry degradation is a key factor for planning and scheduling of maintenance activities to keep the safety and availability of railway in an acceptable level. The aim of the present paper is to predict track geometry condition using the application of Neural Network. For this purpose, a case study has been done on a specific track line, in which longitudinal level is considered as quality indicator of railway track geometry quality. Several neural network models are developed with different number of neurons to find a model with the best performance. The developed model was verified by comparing to inspection geometry data and the results indicate that the developed model using Neural Network is able to properly predict track geometry condition.

railway track geometry, degradation, prediction, Neural Network, maintenance