Influence of Measurement Errors of Input Data on Path Flow Estimations from Probe Vehicle Information and Modification of Estimated Path Flows

Abstract

Total flow and O-D flow based models for estimating path flows using probe vehicle information and traffic counts are constructed as a mathematical programming in which probabilities of simultaneously occurrence are maximized. First, when input data of prior total flow and O-D flows, observed link flows, and a prior probability of each path flow have some errors and inconsistency, differences of two proposed models are analyzed. Secondly, modification methods of input data are proposed to update estimated path flows. The proposed models were applied to a simple network. As a result, link traffic counts and prior O-D flows and total flow had some errors and were inconsistent, the proposed models were able to yield estimations of path flows. Furthermore, it was clarified that accuracy of estimated path flows was improved by using the proposed modification methods.