SIGNAL+DRAHT | Ausgabe 07-08/2012

Ranking of important components in railway systems – comparison of two probabilistic graphical models

August 2012 | Eric Schöne, Jochen Trinckauf, Michael Kunze, Qamar Mahboob, Ulrich Maschek

Component Importance Measure (CIM) can be useful in improving design, op­eration and maintenance of a railway system. This article discussed a number of CIM and their computation using Fault Tree Analysis (FTA) and Bayesian Networks (BNs). Both the methods were able to compute the CIM for a spe­cific safety problem from railway industry. The application of BNs to railway industry is novel and the additional benefits in using BNs are that they model joint distributions (dependencies) of random variables (system components) in a compact and flexible way.