2022
Journal
Sustainability
Abstract
The serially-correlated nature of engine operation is overlooked in the vehicular fuel and emission modelling literature. Furthermore, enabling the calibration and use of time-series models for instrument-independent eco-driving applications requires reliable forecast aggregation procedures. To this end, an ensemble time-series machine-learning methodology is developed using data collected through extensive field experiments on a fleet of 35 vehicles. Among other results, it is found that Long Short-Term Memory (LSTM) architecture is the best fit for capturing dynamic and lagged effects of speed, acceleration, and grade on fuel and emission rates. The developed vehicle-specific ensembles outperformed state-of-the-practice benchmark models by a significant margin and the category-specific models outscored the vehicle-specific sub-models by an average margin of 6%. The results qualify the developed ensembles to work as representatives for vehicle categories and allows them to be utilized in both eco-driving services as well as environmental assessment modules.
2021
Conference
Transportation Research Board 100th Annual Meeting
Abstract
The road transport sector, is one of the main energy consumers and contributors to the emissions leading to climate change. Several models are developed across the globe for environmental assessment of transportation projects. However, validity of using them in regions rather than their origins without major adjustments is questionable. In this study, we validate US EPA’s MOVES model for use in Canada by performing several on-road fuel and emissions measurement experiments in cities from Canada, Colombia, and Iran. State-of-the-technology portable activity and emissions measurement systems are used for data collection. Our findings revealed distinct differences between the ground-truth and MOVES predictions. MOVES under- estimated energy consumption and CO2 rates with mean percentage errors of -17% and -35%, respectively. The model predicts energy consumption and CO2 fairly better for automatic transmission vehicles compared to manual ones, and significantly better for turbo-charged-engine vehicles and the light-weight/small vehicle segments (by R-squared margins of 19%, 51%, and 45%, respectively). For NOx and particulate matters, there is a dramatic over-estimation as the mean percentage error goes up to +420%. The results prove the idea that every region needs its own locally-developed fuel and emissions model as regions are diverse in terms of the fleet characteristics, transportation network topology and control, weather conditions, driving habits, etc.
2020
Journal
Transportation Research Part D: Transport and Environment
Abstract
Many of the existing studies on vehicular fuel consumption estimation are criticized in aspects such as ignoring real-world training data, low diversity of test fleet, impracticality of models in real-world applications (i.e. instrument-independent eco-driving), or their prediction power in the non-linear multi-dimensional space of fuel consumption estimation. In this paper, we proposed a machine learning modelling method using large on-road data collected from a fleet of 27 vehicles. The usability of models in absence of specialized instruments was in focus. We tried to improve the accuracy of our base models by introducing engine-speed estimates through a cascaded modelling procedure. As a result, the accuracy of models reached 83%, while improvements as high as 37% were achieved depending on the technique (support vector regression or artificial neural networks) and vehicle class. Finally, we took the first step from vehicle-specific models towards category-specific modelling by a categorical analysis over fleet attributes.
Conference
Transportation Research Board 99th Annual Meeting
Abstract
As vehicle-miles traveled continue to rise, we must be more vigilant than ever of vehicular emissions. Intersection controls, such as stop signs and traffic lights, are not often thought of as contributors to emissions and are implemented in neighborhoods as simple and low-cost traffic safety solutions. In this research, we use a combination of portable activity and emissions monitoring systems to gather data from selected neighborhoods in Montreal, Quebec, and to evaluate the emissions generated at intersections with the three main types of controls (intersection approach without stop sign, intersection with stop, and with traffic light). 30-meter buffer zones are used as the scope of impact. A growing trend is visible in energy consumption and emission indexes when moving from no-stop intersections to all-way-stops and finally, to the signalized ones. However, when the data is controlled for the time spent within each intersection, the trend disappears. CO2 and fuel consumption rates hardly vary between the intersection types and the NOx emission rates show only a weak trend when moving towards stricter signal controls. We use the AADT of a typical intersection in the study area to calculate the expected growth of emissions annually if such intersection were to be changed from a no stop intersection to a signalized intersection. We observe increases as much as 112% in rates. Finally, we compare the ground-truth data to the outputs generated by MOVES, the EPA’s emissions model. MOVES outcomes were found to be very different from those measured in the field.