Self-adapting traffic performance forecasts using SOM

Research output: Chapter in Book/Report/Conference proceedingConference article in proceedingsScientific


An online short-term prediction model of highway travel time was made using MLP-neural networks on a Finnish two-lane two-way highway. The results of the online application of the model showed that the use of even a simple prediction model could improve the quality of travel time information. However, the problem remained that if the congestion phenomenon changes (even slightly), the neural network needs new training. In addition to this, there is often too little time to collect training data, which leads to a small number of samples that represent random incidents and consequently to a need to learn while working online. The purpose of this ongoing study is to make a self-adapting short-term prediction model of the flow status. At present, the information of the flow status is given on the Internet based on latest measurements of travel time. The objective is to improve the quality of that information.
Original languageEnglish
Title of host publicationGerman-Dutch-Finnish Seminar on traffic engineering
Subtitle of host publicationCD-ROM
Publication statusPublished - 2004
MoE publication typeB3 Non-refereed article in conference proceedings
EventGerman - Dutch - Finnish seminar on traffic engineering - Otaniemi, Espoo, Finland
Duration: 3 Oct 20075 Oct 2007


SeminarGerman - Dutch - Finnish seminar on traffic engineering


Dive into the research topics of 'Self-adapting traffic performance forecasts using SOM'. Together they form a unique fingerprint.

Cite this