Solving Location Problem For Vehicle Identification Sensors To Observe And Estimate Path Flows In Large-Scale Networks

Abstract

Origin-Destination (OD) demand is one of the important requirements in transportation planning. Estimating OD demand could be an expensive and time consuming procedure. These days using vehicle identification sensors for OD estimation has become very common because of its low cost and high accuracy. In this paper, we focus on solving two location problems of these sensors: one to observe and one to estimate path flows. These problems have only been solved for small-scale networks until recently due to being computationally expensive. Therefore, we try to present a method to solve these models for large-scale networks. Due to resemblance of these models and set covering problem, we used heuristic and meta-heuristic methods based on set covering problem. For this purpose, we defined our new set covering matrix based on prime matrix. In order to determine which method is more appropriate, we chose a large-scale and six medium-scale networks. The results represent that through heuristic methods and meta-heuristic methods a greedy algorithm and a Tabu search are more appropriate respectively.

Time-dependent estimation of origin-destination matrices using partial path data and link counts

“…Because ALPR data and actual link vehicle counts (observation data in the time interval 𝑡) are not available, the real OD matrices were assigned to the network to capture detected flows of partial paths and links for each time interval. Yazdi and Shafahi (2018) developed a greedy heuristic algorithm to locate ALPR cameras on the network. They improved Vasko's method (Vasko et al 2016) for solving vehicle identification sensor locations for large-scale networks.…”