The paper describes the procedures for GPS/GPRS data collection with the aim of improving the urban travel time modelling process. Data were collected from multiple sources and then fused based on common attributes. Three primary sources are GPS/GPRS vehicle data, meteorological and road infrastructure data. GPS/GPRS data include log time (time at which the record is generated expressed in UTC), vehicle ID (identification number of the vehicle/GPS device), X coordinate (X coordinate of GPS track), Y coordinate (Y coordinate of GPS track), speed (speed of the vehicle acquired from GPS in km/h), course (angle at which the vehicle is travelling with reference to the North), GPS status (3 values that indicate accuracy) and engine status (indicates whether the vehicle's engine was turned off or on). Meteorological and Hydrological Service data include air and ground temperature, rain, visibility, air pressure, wind, snow etc. and GIS database infrastructure data include road link and sublink IDs as well as length, road category, direction, beginning and end coordinates, name etc. These data were cleansed and the map-matching procedure was carried out. All the data were collected on the same urban area during the same time period for better understanding of transportation activities. The fused data have been analysed and factors have been derived as the main carriers of input information for the travel time modelling in urban areas. The idea is to identify the main elements that affect the travel time and congestion in the urban area so they could be built into the model to achieve a satisfying level of prediction accuracy.

Tehnologija prometa i transport