A hybrid model for speed estimation based on single-loop data

Abstract

The accurate, real-time traffic speed is very important for a successful freeway traffic management system. Unfortunately, widely embed single-loop detectors cannot give the traffic speed directly. A hybrid model for speed estimation based on single-loop data is presented. Different models are used to estimate speed according to different traffic conditions such as free flow and congested flow. The practice test using the data provided by WSDOT gave a good result. The traffic speed estimated by this method is very close to the speed observed by the corresponding dual-loop detector. Compared with the commonly adopted speed estimation algorithm, the proposed method improves estimation accuracy significantly.