Hydrogen fuel cell vehicles (HFCVs) are a promising technology for reducing vehicle emissions and improving energy efficiency. Due to the ongoing evolution of this technology, there is limited comprehensive research and documentation regarding the performance of HFCVs. To address this void, the paper develops a simple HFCV energy consumption model using new fuel cell efficiency estimation methods. The suggested energy efficiency model employs real-time vehicle speed, acceleration, and roadway grade as input parameters for calculating instantaneous hydrogen energy consumption rates, battery energy consumption, and total energy consumption. The results suggest that the model's forecasts align well with real-world data, demonstrating average error rates of 0.0% and -0.1% for fuel cell energy and total energy consumption across all four cycles. However, it was observed that the error rate for a particular cycle, such as the UDDS, can be as high as 13.1%. Moreover, the study confirmed the reliability of the suggested model through validation with independent data. The findings indicated that the proposed model precisely predicted energy consumption, with an error rate of 6.7% for fuel cell estimation and 0.2% for total energy estimation compared to empirical data. Additionally, instantaneous SOC predictions from the model closely match observed instantaneous SOC measurements, highlighting the model's effectiveness in estimating real-time changes in the battery SOC. The study investigated the energy impact of various intersection controls to assess the applicability of the proposed energy model. The simplicity of the proposed model enables easy estimation of energy consumption using traffic simulation models.

In this paper, bikeshare data from Chicago on weather-friendly days in 2019 and 2020 were analyzed to investigate the variation in bikeshare travel before and during the pandemic. Our results show that bikeshare trips during the pandemic were much longer than prior to the pandemic. The increased rate of bikeshare usage was unbalanced spatially and varied significantly for different user types. Bikeshare was used remarkably more by casual users than by subscribers, and the increase occurred much more in the outskirts of the city. The increase in bikeshare travel was associated with a reduction in travel by ridehailing and public transit, especially in the urban periphery. The correlation of bikeshare use with the bus system was much less significant than with the rail system. Bike lanes/facilities had a mixed effect on bikeshare travel. Weekend bike trips increased in areas where there was no bike lane. Weekday trips, on the contrary, increased in the vicinity of bike greenways.

Predicting signal phase and timing (SPaT) most likely switching times and confidence level is needed to enhance green light optimal speed advisory (GLOSA) systems. This study proposes an architecture based on transformer encoders to improve prediction performance. This architecture is combined with different deep learning methods, including Multilayer perceptrons (MLP), long-short-term memory neural networks (LSTM), and Convolutional long-short term memory neural networks (CNNLSTM) to form an ensemble of predictors. The ensemble is used to make data-driven predictions of SPaT information for six different intersections along the Gallows Road corridor in Virginia. The study outlines three primary tasks. Task one is predicting whether a phase would change within 20 seconds. Task two is predicting the exact change time within 20 seconds. Task three is assigning a confidence level to that prediction. The experiments show that the proposed transformer-based architecture outperforms all the previously used deep learning methods for the first two prediction tasks. For the first task, the transformer encoder model provides an average accuracy of 96%. For task two, the transformer encoder models provided an average mean absolute error (MAE) of 1.49 seconds, compared to 1.63 seconds for other models. Consensus between models is shown to be a good leading indicator of confidence in ensemble predictions. The ensemble predictions with the highest level of consensus are within one second of the true value for 90.2% of the time as opposed to those with the lowest confidence level, which are within one second for only 68.4% of the time.

This paper develops a power-based microscopic energy consumption model for battery electric vehicles (BEVs) entitled Virginia Tech Comprehensive Power-based Energy Consumption Model 2.0 (VT-CPEM 2.0). This model avoids bang-bang control, is calibrated using US Environmental Protection Agency (EPA) city and highway energy consumption ratings; along with publicly available vehicle and roadway pavement parameters; and captures regenerative braking. The model validation considers three BEVs and three drive cycles. The test results show that the estimated energy consumption rates by the proposed model are consistent with in-laboratory data collected by the Argonne National Laboratory producing an average error of 3.9% and a maximum error of less than 7.5% for three vehicles across three drive cycles. Additionally, the model demonstrates that the optimal speed of BEVs increases with an increase in the auxiliary power.

This research enhances and evaluates the performance of a Decentralized Nash Bargaining (DNB) adaptive traffic signal controller that operates a flexible National Electrical Manufacturers Association (NEMA) phasing and timing scheme responding dynamically to the fluctuating traffic demand. The DNB controller is enhanced to 1) use traffic density estimates instead of queues to optimize signal timings; 2) to consider the eight-phase two-ring NEMA controller configuration within the game-theoretic approach; and 3) to consider dynamically adaptable control time steps. The enhanced DNB controller is benchmarked against 1) fixed-time traffic signal control using the state-of-the-practice Webster's method and an emerging Laguna-Du-Rakha (LDR) method for computing the optimum cycle length; 2) state-of-practice actuated traffic signal control; and 3) state-of-art Reinforcement Learning (RL) traffic signal control. The controller is tested on two isolated signalized intersections demonstrating enhanced overall intersection performance compared to the baseline pre-timed and actuated controllers at various demand levels and offers comparable performance to a previously developed RL controller. Specifically, the DNB controller results in a decrease in the average vehicle delay and queue size by up to 54% and 63%, respectively compared to Webster's state-of-the-practice pre-timed control. Unlike the RL controller, the DNB controller requires no pre-training while adapting to fluctuating traffic conditions, thereby providing a flexible framework for reducing traffic congestion at signalized intersections. As such, this research contributes to the development of smarter and more responsive urban traffic control systems.

A framework combining data driven deep learning with domain-driven controller logic is introduced in this paper. This is done by encapsulating the controller logic into a digital twin that generates Signal Phasing and Timing (SPaT) information created by the controller given the controller logic and actuation data. The digital twin is designed as an extensible object-oriented program that can run provided the controller data. The controller data are used in two ways. First, it is used to initialize the digital twin by extracting the controller settings from the data either through explicitly mentioned variables in the data or implicit variables inferred from the data. Second, a deep learning multilayer perceptron (MLP) model is used with the data to estimate the actuations. Actuations are then fed to the controller digital twin for the digital twin to provide SPaT predictions. This approach is tested on a single intersection and provides a much better overall prediction performance than other approaches with perfect predictions 80 percent of the time and within 1 second of actual switch times 85 percent of the time. The mean absolute error and mean absolute percentage error in this approach are 0.67s and 7.61 percent respectively, resulting in a more than 58 percent reduction in the prediction error compared to other models presented in the literature.