Freeway Link Travel Time Prediction Research And System Implementation Based On Float Car Data

Link travel time prediction is an important part of the transportation intelligence information prediction. It applies an effective model to predict travel time of some link in next period or next couple of periods. Link travel time prediction is useful for traffic managers and travelers. To increase travel efficiency, travelers can refer to the predicted link travel time before they plan the travel route. Traffic managers can carry out traffic diversion and traffic control in advance according to the predicted link travel time, in order to effectively reduce the traffic congestion. Link travel time prediction requires historical and real-time traffic information. Floating car system is a new traffic information collection method. Floating cars equipped with GPS devices send data back to the terminal periodically when they are driven on the roads. By using these real-time data of floating cars, we can estimate travel time of the links that floating cars are driven through. In this paper, we use floating car data provided by Guangdong Provincial Communications Department to predict travel time of various links of freeways in GuangDong Province.Vehicle location information contained in Floating car data includes only the latitude and longitude, so we need to match the floating car to the road it is traveling. The matching process is known as map matching. This paper designs a simple and effective map matching algorithm based on the characteristics of freeway. After we finish matching the floating cars to the links, some method is used to convert the floating car data to link travel time. We propose a link travel time prediction method combining empirical mode decomposition and support vector regression. We carry experiment with traffic data set from Washington state highway and floating car data provided by Guangdong Provincial Communications Department. Experimental results show the proposed method is effective. Finally, the above-mentioned map-matching algorithm and the link travel time prediction method are applied to implement a visualization system to predict travel time of various links of freeways in GuangDong Province based on ArgisEngine tool.