Research for Increasing FCD Map Matching Accuracy Based on Feature Extraction of Continuous Traffic Flow and Interrupted Traffic Flow

Abstract

Statistics and forecasting of road traffic conditions is a key element for transportation planning. Because of the accuracy limitations of FCD (floating car devices, such as BEIDOU or GPS devices installed on a vehicle), it is very difficult to do map matching and route deriving for elevated and ground road networks when using FCD data. Because of this, the accuracy of road traffic judgment is very low. This study deeply analyzed and mined FCD data obtained from taxi on-board devices for Shanghai city, used grey relational analysis theory and fuzzy pattern recognition theory to extract features of continuous traffic flow and interrupted traffic flow, and improved accuracy for recognizing real traffic conditions. As the result, the recognition rate increased greatly—to more than 90 %.