Traffic flow estimation on the expressway network using toll ticket data

摘要

Traffic flow estimation (TFE) is an important task for intelligent transportation systems, especially for expressway system in travel demand management, road conditions monitor etc. Traffic flow is usually detected by fixed sensors, which is neither practical nor economically feasible to cover all expressway component links. Additionally, the existing research usually ignores the effects of drivers' dwell behaviours on TFE. To fill these gaps, this study proposes a TFE method based on toll ticket data (TTD) which records trip information, except the actual route. Meanwhile, dwell time, as a key component of drivers' journeys, will be integrated into the proposed approach. In this study, a Gaussian mixture model for the origin-destination pairs with multiple alternative routes will be formulated. Moreover, expectation-maximisation algorithm is introduced to estimate link traffic flows and dwell time distributions simultaneously. By TTD collected from the Shandong Province Expressway in China, the proposed approach is calibrated and applied to TFE empirically. The results show that the proposed approach has a good performance on TFE with consideration of dwell time.