Macroscopic Fundamental Diagram Estimation Considering Traffic Flow Condition of Road Network

Abstract

A macroscopic fundamental diagram (MFD) is an important basis for road network research. It describes the functional relationship between the average flow and average density of the road network. We proposed an MFD estimation method based on the traffic flow condition. Firstly, according to statistical theories, the road network data are divided into three traffic flow conditions (free flow, chaotic and congested) bounded by a 95% confidence interval of the maximum traffic capacity of each intersection in the road network. Then, in each condition, we combined principal component analysis and the Jolliffe B4 method to reduce dimension for extracting critical intersections. Finally, the full-scale dataset of the road network was reconstructed to estimate the road network MFD. Through numerical simulation and empirical research, it is found that the root mean square error and absolute percentage error between estimated MFD and true MFD considering the traffic flow condition are smaller than those without considering the traffic flow condition. The MFD estimation and the division of the traffic states of the road network were completed at the same time. The proposed method effectively saves the time cost of road network research and is highly accurate.

Keywords

• macroscopic fundamental diagram
• traffic state
• confidence interval
• principal component analysis
• jolliffe b4 method