Fundamental Diagram Modelling and Analysis Based NGSIM Data

Abstract

Abstract Traffic control requires looking at traffic models of all types in finer details. This paper is to investigate lane-wise flow-density (or equivalently speed-density) relationship which is traditionally called Fundamental Diagram (FD) over a stretch of homogeneous freeway section using the microscopic NGSIM data. Particularly, it investigates how a homogenous traffic further drop (breakdown) through data analysis and modeling. The breakdown of a homogenous traffic is understood as the significant flow drop and density increase with noticeable shock-wave back-propagation. The corresponding density is a generalization of the critical density for traffic breakdown from free-flow. Variable structure models with two limbs of the inverse λ – shape are proposed to model the homogenous flow and its further drop. A special Generalized Polynomial Model (with fraction coefficients) is also proposed for the right limb. Properly aggregated NGSIM data are used to fit the model with results compared with some other models over time at fixed location using Root Mean Square Errors (RMSE) as measure. Principles for time/distance aggregation of individual vehicle trajectories to macroscopic traffic state parameters are proposed.