Machine state classification of electric track circuit by means of logistic regression

Abstract

Electric track circuits are widely used on railways as sensors providing position of a train and information about physical integrity of rails. A modern railway monitoring system is required to have automatic data analysis capabilities. For a track circuit this functionality can be implemented as automatic state classification. To perform this task, we developed an algorithm based on logistic regression. In this article we describe basic principles of the algorithm and machine learning techniques that are applied.