Link-level Travel Time Prediction Using Artificial Neural Network Models

Mane, Ajinkya S.; Pulugurtha, Srinivas S.

doi:10.1109/itsc.2018.8569731

Cited by 8 publications

(5 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Travel patterns are governed by the area type, socioeconomic characteristics, and land use categories of surrounding land (27)(28)(29)(30)(31). These components of urban areas assist in identifying planning strategies to construct new facilities or implement transportation projects.…”

Section: Literature Reviewmentioning

confidence: 99%

“…Some of the most common short-term forecasting approaches include time-series estimation ( 16 – 18 ) and regression-based statistical analysis ( 19 , 20 ). Other approaches, like artificial intelligence techniques (neural networks), have also been explored in the past to predict travel time ( 21 – 27 ).…”

Section: Literature Reviewmentioning

confidence: 99%

“…Like the travel demand in an area, the travel time patterns of a road vary based on the facility type, the surrounding conditions, and the area type. Further, the travel time of a road link is influenced by the land use and socioeconomic characteristics of the surrounding area ( 27 , 31 ). The complex interactions between the surrounding area characteristics and the travel patterns require a research design that considers the spatial parameters.…”

Section: Literature Reviewmentioning

confidence: 99%

See 2 more Smart Citations

Modeling Road Link-Level Truck Travel Times Using On- and Off-Network Characteristics

Duvvuri

Pulugurtha

2022

Transportation Research Record

Self Cite

View full text Add to dashboard Cite

Travel times of freight trucks (or trucks) play a major role in trip planning, identification of efficient routes, and allocating resources for implementation of strategies. While research on travel time estimation models for passenger cars or traffic stream is documented in the literature, their applicability for trucks remains debatable. Truck travel is influenced by the road characteristics, surrounding land use, and demographics of an area. The focus of this research, therefore, is on estimating the truck travel times using the on-network (road) and off-network (land use and demographics) characteristics. Truck travel time data for 501 road links in North Carolina from 2019 were processed for four times of the day and two days of the week. Generalized estimating equations (GEE) were used with the average truck travel time per mile (ATTTPM) as the dependent variable. Spatial proximity (buffers of 0.25 mi, 0.50 mi, and 1 mi) and spatial weights (distance decay functions like 1/d, 1/d2, and 1/d3) were explored to check the best possible approach for capturing off-network characteristics. The office (business park and administrative areas), transportation (rest areas and parking facilities), heavy commercial, and light industrial land uses have an increasing influence on the ATTTPM. The linear model developed using data from a 0.25-mi buffer is best suited to estimate the ATTTPM. Off-peak hour delivery incentives or truck priority systems on road links near these land uses can be implemented to improve truck mobility. The methodology illustrated in this research is transferable and could be used for estimating truck travel times across a region.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

Section: Literature Reviewmentioning

confidence: 99%

Section: Literature Reviewmentioning

confidence: 99%

See 1 more Smart Citation

Modeling Road Link-Level Truck Travel Times Using On- and Off-Network Characteristics

Duvvuri

Pulugurtha

2022

Transportation Research Record

Self Cite

View full text Add to dashboard Cite

show abstract

“…Linear modelling (X. Zhang and Rice, 2003), the nonlinear autoregressive with external inputs (NARX) model (Mane and Pulugurtha, 2018), the nonlinear autoregressive model (NAR) (Mane and Pulugurtha, 2018), clustering (Elhenawy, H. Chen, and Rakha, 2014), neural networks (Mane and Pulugurtha, 2018), ant colony based approach , and deep neural networks (Duan, Lv, and F.-Y. Wang, 2016;Ran et al, 2019).…”

Section: Literaturementioning

confidence: 99%

Advancements in multi-objective eco-routing solutions using connectivity

Alfaseeh¹

2021

Preprint

View full text Add to dashboard Cite

Due to the significant adverse impact of transportation systems on the environment, topics related to alleviating greenhouse gas (GHG) emissions are gaining more attention. As potential solutions to mitigate GHG emissions, several approaches have been proposed to better control traffic and manage transportation systems. The employment of Intelligent Transportation System (ITS), which adopts the advancements in Information and Communication Technology (ICT), has been proposed as the most favourable approach to alleviate the undesirable impact of transportation systems on the environment. ITS can control several aspects of a network, such as speed, traffic signals, and route guidance. For the purpose of routing, this research aims to exploit the advancements in ICT by including connected and automated vehicles (CAVs) and sensing technology in an urban congested network.<div>Anticipatory multi-objective eco-routing in a distributed routing framework was proposed and compared to myopic routing with a large case study on a congested network. The End-to-End Connected Autonomous Vehicles (E2ECAV) dynamic distributed routing framework was examined, and encouraging results were found based on the traffic and environmental perspectives. The impact of different market penetration rates (MPRs) of CAVs was examined for various traffic conditions. E2ECAV was adopted for both the myopic and anticipatory routing strategies in this dissertation. The best GHG costing approach was defined and was among the elements tackled in this research. For a robust anticipatory routing application, predictive models were developed based on Long-Short Term Memory (LSTM), a deep learning approach, while considering a high level of spatial (link level) and temporal (one minute) resolution. With regards to the LSTM predictive models, the impact was illustrated of using a deeper LSTM network and systematically tuning its hyper-parameters. The anticipatory routing strategy significantly outperformed the myopic routing strategy based on the the traffic and environmental perspectives. This research shows that ITS can help significantly reduce GHG emissions produced by transportation systems. The developed predictive models can be used while real-time data are collected from sensors within an urban network. Furthermore, the proposed anticipatory routing framework can be applied in a real-time situation. </div>

show abstract

“…For directly forecasting travel time, several predictive models have been employed. Linear modeling (Zhang and Rice, 2003), nonlinear autoregressive with external inputs (NARX) model (Mane and Pulugurtha, 2018), nonlinear autoregressive model (NAR) (Mane and Pulugurtha, 2018), neural networks (NNs) (Mane and Pulugurtha, 2018), and deep neural networks (Duan et al, 2016;Ran et al, 2019). On the other hand, a large body of literature exists where travel time was implicitly predicted from speed, including Ma et al (2015), Yao et al (2017), and Gu et al (2019).…”

Section: Introductionmentioning

confidence: 99%

Deep Learning Based Proactive Multi-Objective Eco-Routing Strategies for Connected and Automated Vehicles

Alfaseeh¹,

Farooq²

2020

Front. Future Transp.

View full text Add to dashboard Cite

This study exploited the advancements in information and communication technology (ICT), connected and automated vehicles (CAVs), and sensing to develop proactive multi-objective eco-routing strategies for travel time and Greenhouse Gas (GHG) emissions reduction on urban road networks. For a robust application, several GHG costing approaches were examined. The predictive models for link level traffic and emission states were developed using the long short-term memory (LSTM) deep network with exogenous predictors. It was found that proactive routing strategies outperformed the reactive strategies regardless of the routing objective. Whether reactive or proactive, the multi-objective routing, with travel time and GHG minimization, outperformed the single objective routing strategies. Using a proactive multi-objective (travel time and GHG) routing strategy, we observed a reduction in average travel time (17%), average vehicle kilometer traveled (22%), total GHG (18%), and total nitrogen oxide (20%) when compared with the reactive single-objective (travel time).

show abstract

Link-level Travel Time Prediction Using Artificial Neural Network Models

Cited by 8 publications

References 24 publications

Modeling Road Link-Level Truck Travel Times Using On- and Off-Network Characteristics

Modeling Road Link-Level Truck Travel Times Using On- and Off-Network Characteristics

Advancements in multi-objective eco-routing solutions using connectivity

Deep Learning Based Proactive Multi-Objective Eco-Routing Strategies for Connected and Automated Vehicles

Contact Info

Product

Resources

About