AUTOMATIC MULTIREGIME FUNDAMENTAL DIAGRAM CALIBRATION USING LIKELIHOOD ESTIMATION

Abstract

LIMB (Likelihood Identification for Multi-regime models’ Breakpoint), a new tool, is developed to calibrate various Fundamental diagrams (FDs) under different road geometric and traffic operational conditions. This tool enables to estimate traffic state from real time data and is ready to implement into model based control strategy. Since efficient traffic management or control of transportation system remains big challenge due to the necessity of accurate traffic state estimation; model based control strategy incorporated with LIMB tool can ameliorate the complexity. This research found that the breakpoint of multi-regime FD models obtained from experience were not able to estimate traffic state precisely; therefore LIMB was used to calibrate those models. The investigation also endeavored to develop a guideline which was capable to calibrate suitable FD models for lanewise traffic conditions. Our proposed technique is independent of speed limits and completely automatic without any threshold inputs. Furthermore, it is comparable with the well-recognized FD automatic calibration technique. The comparative study found a 5% to 8% variation in estimating FD parameters. Later, we investigated several novel single and multiregime FD models utilizing field traffic data obtained from PeMS website. LIMB adopted likelihood estimation method to identify density at breakpoint in between free flow and congestion states for multi-regime models. It applies least square method to estimate critical density-free flow speed-capacity. The proposed interface is conducive and easily adaptable for transportation practitioners to select the best model based control strategy for smooth and efficient traffic operation.