Semi-dynamic Markovian path flow estimator considering the inconsistencies of traffic counts

This paper proposes a semi-dynamic path flow estimator (PFE) that does not require enumeration of the paths set. The proposed model takes traffic assignments derived using the absorbing Markov chain approach and incorporates them into the nonlinear PFE model while considering the inconsistency of traffic counts described by Chen et al. (2009). The proposed model was applied to two types of toy networks. As a result, we confirmed that the proposed model can estimate the link volume with high accuracy if the observed link volumes are error-free. We further confirmed that the total origin–destination volumes as well as link volumes are underestimated if the observation interval of the traffic counts is longer than the length of the time period for the estimation.