Study On Accuracy Testing And Improvement Of Real-time Traffic Information Based On Floating Car Data

Data collection technology via has been well developed recentlty. As a main source for data collection, floating car owns its good characteristics. The essence of this technology is using samples to reflect the characteristic of population, requiring that enough floating cars should be collected to cover most of roads in each time section.It is a challenge task due to felixable routes of drivers and problems existing in data collection and processing. Besides, the status of one single vehicle may be influenced by various factors, and variation exists between different vehicles. So it is necessary for us to validate the accuracy of real-time traffic information.Firstly, this paper analyses two of main factors affecting vehicle speed-road condition, signalized intersection and v/c, using the data from investigation and simulation to validate the theory. Based on the analysis above, a new method which mainly focuses on merging established subsections is proposed to reduce the variation; also a recommendatble value of subsections from simulation by using clustering analysis is presented. At last, mumerial examples of floating car data from aterials in Shanghai are presented by this new method, the result shows its effectivity in improving traffic information accuracy.