A Systematic Study on Forecasting of Traffic Flows with Artificial Neural Networks

Konferenz: ARCS 2015 - 28th International Conference on Architecture of Computing Systems
24.03.2015 - 27.03.2015 in Porto, Portugal

Tagungsband: ARCS 2015

Seiten: 8Sprache: EnglischTyp: PDF

Persönliche VDE-Mitglieder erhalten auf diesen Artikel 10% Rabatt

Sommer, Matthias; Tomforde, Sven; Haehner, Joerg (Organic Computing, University of Augsburg, Eichleitnerstr. 30, 86159 Augsburg, Germany)

Traffic flow is highly dynamic and complex to foresee, therefore it offers an interesting application domain for Organic Computing. Most traffic management systems try to adapt their traffic signalisation to the current traffic flow patterns, but for an optimal and fast adaptation, traffic flow forecasts are needed. A resilient traffic management system needs the ability to forecast traffic flows in order to pro-actively adapt the signalisation with the goal to decrease or even prevent negative impacts on the traffic network. Artificial Neural Networks have shown to be a powerful tool in forecasting traffic flows. This paper investigates a systematic study of Artificial Neural Networks and presents which variants and parameter settings are most profitable in which situations.