Overhead equipment turnouts fitness assessment by supervised learning

Abstract This paper proposes an optimization technique to make the system more reliable by predicting failures relating to OHE turnouts (like panto entanglement) before they might happen and thereby saving a lot of resources and time. The turnout OHE has been modelled using the various maintenance parameters based on their standard values. Concerning to this, a supervised learning approach that uses optimal features selected by diversity assessment has proposed. The critical objective of the proposal is handling crux of dimensionality raised due to the voluminous input data. The other objective of the proposal is to perform binary classification with minimal false alarming using evolutionary search technique called cuckoo search. The permissible limits are also taken into consideration while modelling the system. The case study has been done on the data collected from Secunderabad division of South-Central Railway.