A Robust Framework for the Estimation of Dynamic OD Trip Matrices for Reliable Traffic Management

Barceló, Jaume; Montero, Lídia · 2015 · Crossref

DOI: 10.1016/j.trpro.2015.09.063

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the challenge of estimating dynamic Origin-Destination (OD) trip matrices, which are essential inputs for traffic assignment and management but are not directly observable. While traditional methods adjust static matrices using link flow counts, dynamic estimation typically relies on space-state formulations like Kalman Filtering. The authors note that the efficiency of real-time estimation heavily depends on the quality of the initial "seed" matrix and the detection layout. Motivated by the availability of Information and Communications Technologies (ICT) data, such as Bluetooth and Wi-Fi sensors, the study proposes an integrated computational framework that combines offline historical data processing with online real-time estimation to improve robustness and accuracy. The proposed framework consists of two main components. First, an offline procedure generates time-sliced OD matrices using historical traffic data, weather information, and calendar events. This process employs a Static Bilevel OD Adjustment Procedure combined with a Dynamic User Equilibrium (DUE) assignment based on mesoscopic traffic simulation to identify Most Likely Used Paths (MLU). These adjusted matrices serve as high-quality seeds for the second component: an online real-time estimator. This estimator utilizes a linear Kalman Filter variant (KFX3) that treats deviations of OD path flows as state variables. Crucially, the model incorporates real-time travel time measurements from ICT sensors to update time-varying model parameters, thereby eliminating the need for nonlinear approximations of traffic dynamics. The system is designed to handle variable sensor configurations, including detector malfunctions, by automatically adapting to active sensors. The study validates the framework using a parallel highway network model from Hu et al. (2009). The experimental design involves a network with four OD pairs and ten MLU paths, monitored by ICT sensors at entry nodes and intersections. The simulation assumes a one-hour horizon divided into 15-minute time slices, with all vehicles assumed to be ICT-equipped for verification purposes. The Kalman Filter formulation uses discrete approximations of travel time distributions derived from sampled ICT data, updating these distributions every time subinterval to account for congestion and temporal dispersion. The model explicitly handles conservation equations at entry points and maps state variables to observations through linear transformations involving time-varying fraction matrices. The significance of this work lies in its unified approach to OD estimation, which mitigates the sensitivity of real-time estimators to poor initial seeds by ensuring high-quality offline inputs. By leveraging ICT data to measure travel times directly, the framework simplifies the mathematical formulation from nonlinear to linear, enhancing computational efficiency and reliability. The results demonstrate that the integrated architecture can robustly estimate dynamic OD matrices even when detection integrity is compromised by sensor failures. This approach provides a scalable solution for Advanced Transport Management Systems, allowing for more accurate short-term traffic predictions and management by effectively combining historical patterns with real-time ICT measurements.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• exposure measurement