MINING DATA FROM INDUCT LOOPS

The complicated traffic situation in Prague is not only due to a rise of the number of cars in the last few years, but also due to historical formed streets which are not able to fully satisfy present traffic performances. Also the ratio of motorization is higher here than, for example, in Vienna or Cologne. Using standard methods cannot solve the enormous traffic problems - a new approach should be used. The traffic control in an area of Prague discussed in this paper is based on an adaptive control MOTION. Often, the level of service (LOS) in this area is on the threshold of saturation when the control system cannot significantly improve the situation. For this reason it has to be supported by another system, which provides information about the current traffic conditions to drivers and together create a hybrid traffic management system. Variable message signs (VMS) are used to distribute this information. This paper describes the way how information about LOS can be collected from the induct loops and distributed to the VMS. Most attention is paid to an automated mining of data from detectors. The algorithm used is based on clustering using genetic algorithms. traffic management office, from the police or some random sources (such as drivers passing through critical parts of the system) or from traffic detectors. Probably the most common detectors in most cities are so called induct loops. The data on each detector are collected in regular time intervals and their amount is really huge. In many cases, the data are collected but no algorithm is known to determine which data are significant and how they can be used. In this paper, we would like to suggest a solution on how to choose relevant information from these detectors. We will provide an overview of the steps that must be accomplished in order to extract data from induct loops. Usually the VMS display information about the level of service (LOS) or about travel time. In this paper we focus on methodology on how to determine the level of service.