Author(s):
1. Miloš Simonović, Faculty of Mechanical Engineering in Niš, Serbia
2. Emina Petrovic, Faculty of Mechanical Engineering in Niš, Serbia
3. Vlastimir Nikolic, Faculty of Mechanical Engineering in Niš, Serbia
4. Ivan Ćirić, Faculty of Mechanical Engineering in Niš, Serbia
5. Aleksandar Miltenović, Faculty of Mechanical Engineering in Niš, Serbia

Abstract:
Marshalling yards play important role in freight railway transport. The efficient use of marshalling yards has a deep impact on the efficiency and reliability of rail freight services due to reduction of transportation cost and increasing reliability and punctuallity. Main processes of marshalling yards may take 10–50% of total train transit time. Marshalling processes still involve much manual planning and improved decision support and analysis tools have shown to have great potential. A novel yard management IT system should be based on yard scalable model which will enable heuristic and meta-heuristic optimization with current yard resources and powered by machine learning based algorithm that will enable the real time planning and disposition and support decision making. Concept solution of modern intelligent marshallling yard management system is decribed. Optimization of marshalling process done by heuristics and meta-heuristics optimization methods will not be able to solve the problem in real-time. For that reason, a novel method for real time optimization has to be developed. One of the solutions can use the optimization results as a training data for machine learning decision system. The trained machine learning system will then give the optimal, or near-optimal solution of marshalling operations in real time.

Key words:
Marshalling yard, Heuristic optimization, Machine learning, Advanced optimization algorithms

Thematic field:
Mechatronics

Date of abstract submission:
13.03.2017.

Conference:
13th International Conference on Accomplishments in Mechanical and Industrial Engineering

Paper for reviews