Design of Experiment for the Validation of ATP/ATC Odometry Algorithms

The safety and efficiency of a railway control systems depends on many factors, one of these factors being the precision of the train position and the speed estimation. The knowledge of the estimation precision is important in the developing phase of the procedure and in the integration with the automatic train protection system, in order to properly assess the safety margins used to define the system intervention. The estimation is performed by the so-called odometry system, which includes a series of sensors and a software procedure. The data set used to test the algorithm has to represent all the real conditions that will be met in normal conditions. The criteria used to define the set of tests is based on the Design of Experiment (DOE) technique, which allows one to obtain a compromise between the number of needed tests and the complete reproduction of possible cases. Once the set of tests are defined, they can be reproduced on a test rig, like the MI6 rig, developed and realized by Trenitalia. The parameters that mainly influence the algorithm precision are the wheel/rail adhesion, the type of train, the type of operation and the gradient of the line. For each parameter a series of classes has been defined and combined in order to obtain a set of tests able to represent the behavior of the algorithm in all the cases it could meet in reality and then to obtain a reliable estimation of its prevision. In this work the criteria used to define the data set will be described. Then this set will be used to test the odometry algorithm developed for the SCMT, the Italian Automatic Train Protection (ATP) system and the results will be presented in this paper.

Language

  • English

Media Info

  • Media Type: Print
  • Features: Figures; References; Tables;
  • Pagination: 449-458
  • Monograph Title: Computers in Railways X. Computer System Design and Operation in the Railway and Other Transit Systems

Subject/Index Terms

Filing Info

  • Accession Number: 01035766
  • Record Type: Publication
  • ISBN: 1845641779
  • Files: TRIS
  • Created Date: Oct 20 2006 1:49PM