Fakultät für Informatik
Technische Universität München
Multistage Methods for Freight Train Classification
In this work we establish a consistent encoding of freight train classification methods for a hump yard. This encoding scheme presents a powerful tool for efficient presentation and analysis of classification methods, which we successfully apply to illustrate the most relevant historic results from a more theoretical point of view. We analyze their performance precisely and develop new classification methods making use of the inherent optimality condition of the encoding. We conclude with deriving optimal algorithms and complexity results for restricted real-world settings.
Joint work with Peter Marton, Jens Maue, and Marc Nunkesser