A. James Clark School of Engineering
Permanent URI for this communityhttp://hdl.handle.net/1903/1654
The collections in this community comprise faculty research works, as well as graduate theses and dissertations.
Browse
2 results
Search Results
Item NATIONAL ORIGIN-DESTINATION TRUCK FLOW ESTIMATION USING PASSIVE GPS DATA(2023) Sun, Qianqian; Schonfeld, Paul; Civil Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Truck travel estimation plays an essential role in the transportation field. Nationwide truck flows are particularly important for capturing long-distance truck travels. For the estimation at such a scale, the traditional way of conducting surveys is very costly and cumbersome. Nowadays, GPS data are getting popular for supporting transportation studies, with advantages of freshness, cost-effectiveness, real-world representation, high spatial-temporal coverage and resolution. Hence, utilizing GPS data as an alternative data source is worth investigating. This study proposes a comprehensive framework for achieving large-scale truck flow estimation from passive GPS data, with the United States as a study case. This study enriches the research on GPS-based travel estimation and particularly achieves the estimation at a scale as large as the United States for the first time using GPS data. The framework begins with thorough data preparation, in which an enhanced algorithm is designed for removing data oscillations. Then, truck type classification by weight class is conducted through a random forest (RF) algorithm, which enriches GPS-based vehicle classification research. The estimation is by truck type, which provides unique travel patterns by truck type. Then, a comparative trip identification by truck type is conducted and the algorithm’s robustness for such identification is investigated. Finally, an innovative weighting algorithm that integrates reinforcement learning and iterative origin-destination matrix estimation (ODME) is designed to weight the sample truck traffic according to the U.S. truck traffic population level and to mitigate the spatial bias of sample GPS data. Nationwide truck flow estimation is achieved. The results’ reasonableness is discussed from multiple aspects, such as ODME accuracy, spatiotemporal biases, distance distribution, OD distribution, vehicle miles traveled, and interstate OD pairs from selected states. The products obtained from the framework are useful for many transportation studies, such as planning and operation, safety, transportation and environment, and policies. The framework not only enables large-scale truck flow estimation but also yields good accuracy and does not require excessive computation cost. It is straightforward and has a high generalizability for studies of various scales and areas. It should be widely applicable for serving transportation research and practice needs.Item NATIONWIDE ANNUAL AVERAGE DAILY TRAFFIC (AADT) ESTIMATION ON NON-FEDERAL AID SYSTEM (NFAS) ROADS BY MACHINE LEARNING WITH DATA MINING OF BUILT-IN ENVIRONMENT(2020) Sun, Qianqian; Zhang, Lei; Civil Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)This study aims to address the nationwide gap in AADT data on NFAS roads in U.S. With a Spatial Autoregressive Model as a benchmark, two machine-learning approaches, i.e. Artificial Neural Network and Random Forest, show notable improvement in the accuracy of estimating AADT according to five measures, i.e. MSE, RSQ, RMSE, MAE, and MAPE. A data-mining of the built-in environment from three perspectives, i.e. on-road and off-road features, network centralities, and neighboring influences, paves the way for AADT estimation, which covers 87 variables in centrality, neighboring traffic, demographics, employment, land-use diversity, road network density, urban design, destination accessibility, etc. Data integration using different buffering sizes and statistical analysis of linearity and monotonicity promote the variable selection for estimation. When implementing the machine-learning approaches, not only the estimation performance is analyzed, but also the relationship between each variable and AADT, the interplays among variables, variable importance measures are thoroughly discussed.