Utilizing Machine Learning To Cross-Check Traffic Data and Understand Urban Mobility
-
2021-03-01
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Publication/ Report Number:
-
Resource Type:
-
Geographical Coverage:
-
Edition:Final Sept 2019 to March 2021
-
Corporate Publisher:
-
Abstract:At UDOT, PeMS stores point data collected by roadside radar sensors, loop detectors, and micro-loops, while ClearGuide contains statewide probe data. PeMS data can greatly support mobility pattern studies but are only available at detector locations. In contrast, ClearGuide can provide statewide traffic speed information based on probe vehicle data. Hence, the speed estimates in ClearGuide have a high potential to be biased due to the low probe penetration rate. Comparisons between PeMS and ClearGuide data reveal significant differences in 5-min speed, which further prove the data bias and inaccuracy in ClearGuide. However, as ClearGuide can provide statewide traffic information, it has been used to support many traffic operation tasks when PeMS data are not available. Thus, the lack of correcting ClearGuide data can result in unreliable inputs and consequently the failure of traffic operational activities. To tackle this issue, this research aims to develop a set of machine learning models to integrate these two data sources, mitigate data variations, and produce more reliable estimations of the statewide traffic patterns. The research objectives are achieved by two steps. The first step utilizes regression machine learning algorithms to estimate traffic state using probe vehicle and sensor detector data. Also, performance of those machine learning algorithms is compared using a novel estimation framework. The second step aims to develop a hybrid machine learning approach, by creating a new training variable based on the second-order traffic flow model, to improve the accuracy of traffic state estimation.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
File Type:
Supporting Files
-
No Additional Files
More +