Estimating multi-year24/7origin-destination demand using high-granular multi-source traffic data

Abstract

Dynamic origin-destination (OD) demand is central to transportation system modeling and analysis. The dynamic OD demand estimation problem (DODE) has been studied for decades, most of which solve the DODE problem on a typical day or several typical hours. There is a lack of methods that estimate high-resolution dynamic OD demand for a sequence of many consecutive days over several years (referred to as 24/7 OD in this research). Having multi-year 24/7 OD demand would allow a better understanding of characteristics of dynamic OD demands and their evolution/trends over the past few years, a critical input for modeling transportation system evolution and reliability. This paper presents a data-driven framework that estimates day-to-day dynamic OD using high-granular traffic counts and speed data collected over many years. The proposed framework statistically clusters daily traffic data into typical traffic patterns using t-Distributed Stochastic Neighbor Embedding (t-SNE) and k-means methods. A GPU-based stochastic projected gradient descent method is proposed to efficiently solve the multi-year 24/7 DODE problem. It is demonstrated that the new method efficiently estimates the 5-minute dynamic OD demand for every single day from 2014 to 2016 on I-5 and SR-99 in the Sacramento region. The resultant multi-year 24/7 dynamic OD demand reveals the daily, weekly, monthly, seasonal and yearly change in travel demand in a region, implying intriguing demand characteristics over the years.

Data-driven network loading

“…Moreover, the calibration of the infrastructure parameters reflecting the supply both in the link and the node model is often a demanding and time-consuming process which relies on data availability and reliability. While endogenous congestion in DNL is crucial for long-term planning, several researchers (Cascetta, Inaudi, and Marquis 1993;Ashok and Ben-Akiva 2002;Sohn and Kim 2008;Seo and Kusakabe 2015;Nam et al 2017;Ma and Qian 2018) suggest that congestion can instead be given exogenous for short-term planning or monitoring purposes. The later may concern applications where the estimation of historical traffic flows is required, such as emission estimation or OD demand calibration.…”

“…The authors of a more recent study (Ma and Qian 2018) also adopt the continuous packet approach. They derive the travel times of the head and the tail from a link TTF.…”

“…Most of the data-driven flow propagation approaches neglect discussions on FIFO and spill back; principles which are widely investigated in the traditional endogenous DNL models. Furthermore, such data-driven techniques (Ashok and Ben-Akiva 2002;Sohn and Kim 2008;Ma and Qian 2018;Krishnakumari et al 2019) provide limited references on the required field data type and availability.…”

“…Our aim is to accurately model the flow propagation, solely relying on exogenous variables, such as travel time observations. To the best of our knowledge, although several data-driven network loading techniques have been lately suggested (Yang, Lu, and Hao 2017;Ma and Qian 2018;Krishnakumari et al 2019), none of them is developed in compliance with the fundamental traffic flow theory principles. Principles such as FIFO, vehicle conservation and queue propagation are not extensively discussed, while the accuracy of such techniques under congested conditions is not evaluated.…”

Spatial and Temporal Analyses for Estimation of Origin-Destination Demands by Time of Day Over Year

“…It can be seen from this table that most existing models on OD estimation problems either include the DTA process or linkpath indicator except the works by Ma and Zhen [32]. In Ma and Zhen [32], the authors presented a data-driven framework that estimates day-to-day dynamic OD using high-granular traffic counts and speed data collected over many years based on the statistical equilibrium assumption [33], [34]. The proposed framework statistically clusters daily traffic data into typical traffic patterns using t-Distributed Stochastic Neighbor Embedding (t-SNE) and k-means methods.…”