Combining simulation and machine learning approaches for improved time prediction in intermodal transport

Simulation and machine learning share a similar goal of predicting the behavior of a system through data analysis and mathematical modeling. While simulations map reality to analyze systems, whose behavior is too complex for theoretical, formula-based approaches, machine learning uses real-world data as input and output parameters. Simulations are used for many practical problems, e.g. traffic flow, supply chain behavior. The utilization from both approaches enables an improved solution. In this paper, we combine data-driven simulation and supervised machine learning in a hybrid approach. We investigate the added value of machine learning in a simulation for better time prediction and higher accuracy. We do this in the environment of combined passenger and freight transport in intermodal transport chains. In this setting, there are fixed trips and fluctuating capacities. We develop and evaluate our solution in the context of an intermodal transport chain with the CargoSurfer project. In our research, we show the value that machine learning provides for supply chain simulations with different capacity options, load fluctuations, and day-by-day simulation.

Falk joined the KLU as a Ph.D. candidate in September 2019 under the supervision of Prof. Dr. André Ludwig and Prof. Dr. Henrik Leopold. He is conducting research at the intersection of supply chain management and information systems. Falk received his Master’s degree in Information Systems from the University of Bamberg in 2018. In his master thesis, he empirically investigated the adoption of cryptocurrencies and user resistance behavior. Falk completed his Bachelor’s degree in Information Systems at the FOM University of Applied Sciences for Economics and Management in Nuremberg with a specialization in web engineering.