Freeway Segment Speed Estimation Model Based on Distribution Features of Floating-Car Data

In order to gain good performance, the typical speed estimation models based on GPS data require high-sampling-rate GPS data. Due to the insufficient sample amount, these models may become less effective for points with low-sampling-rate GPS data. Therefore, in this study, it is aimed to fill this gap by developing effective speed estimation models for low-sampling-rate GPS data. The distribution features of floating car data (FCD) on the target segment are analyzed, and thereby corresponding segment speed estimation models based on these features are established. The distribution features of FCD on the target segment can generally be divided into three situations. Thus, this study puts forward a self-adaptive algorithm based on three speed estimation models including speed-time integral model, vehicle tracking model and speed-distance integral model to estimate the segment speed. A simulation experiment is conducted with the use of real OD data collected from Guangshen (GS) freeway in China, and the error range of the self-adaptive algorithm in different sample sizes of FCD is analyzed.