The surge in interest in neural networks is justified by their flexibility in learning very complex processes even with a high degree of nonlinearity. In this regard, the field of transport, which shows a large variety of problems, is particularly appropriate for their application. Neural networks are suggested for the reconstruction of flow-density and speed-density relationships on a single motorway section by using detected data. Using the interpolation capability of neural networks it is possible to differentiate these relationships according to weather conditions and to features of traffic flow. From these relationships, worked out for each detection section, it is possible to identify flow stability and instability, by using the criterion of the sign of the first derivative of function. This structural information is used for a successive network which is capable of a real-time classification of flow conditions. The data base was built up on the basis of the data collected over the Italian motorway running between Padua and Mestre. These data are suitable for the prediction because they provide a sample of time and space, along with the collection of data on weather conditions.
Neural-network models for classification and forecasting of freeway traffic flow stability
Modelle mit neuronalen Netzen für die Klassifikation und Vorhersage der Stabilität der Autobahnverkehrsdichte
Control Engineering Practice ; 4 , 2 ; 153-164
1996
12 Seiten, 13 Bilder, 1 Tabelle, 40 Quellen
Aufsatz (Zeitschrift)
Englisch
Neural Network Models for Classification and Forecasting of Freeway Traffic Flow Stability
British Library Conference Proceedings | 1995
|Multiple-Interval Freeway Traffic Flow Forecasting
Transportation Research Record | 1996
|Multiple-Interval Freeway Traffic Flow Forecasting
Online Contents | 1996
|Multiple-Interval Freeway Traffic Flow Forecasting
British Library Conference Proceedings | 1996
|Multiple-Interval Freeway Traffic Flow Forecasting
Online Contents | 1996
|