Fuel-Saving Behavior for Multi-Vehicle Systems: Analysis, Modeling, and Control

Date of Award

8-6-2017

Document Type

Dissertation

Degree Name

Doctor of Philosophy (Ph.D.)

Institution Granting Degree

The Ohio State University

Cedarville University School or Department

Engineering and Computer Science

First Advisor

Ümit Özguner

Second Advisor

Keith Redmill

Third Advisor

Andrea Serrani

Keywords

Controls, multi-agent systems, fufel savings, intelligent transportation, dynamic eco-driving, swarm modeling, traffic simulation

Abstract

Dynamic eco-driving is an umbrella term describing speed control schemes that utilize connected and automated vehicle technology for the purpose of saving fuel. If dynamic eco-driving is to be widely prescribed as an integral part of widespread fuel-saving endeavors, its expected performance as part of the overall traffic system must be analyzed. Specifically, it must be determined to what extent this type of control remains effective in the presence of dense traffic. We present first a series of multi-vehicle traffic simulations which begin to answer important questions surrounding the effects of dynamic eco-driving on traffic and its potential for fuel savings in a mixed traffic environment. Three representative methods of dynamic eco-driving are tested in traffic scenarios and the estimated fuel economy, trip time, and average speed results are compared. Independent variables include technology penetration rate and amount of traffic, quantified by the delay level of service of the road network's traffic light facility. It is shown that, for the given test cases, average mpg increases linearly with technology penetration rate and dynamic eco-driving causes an increase in average mpg regardless of traffic amount. Overall, results are promising for the usefulness of this clever class of fuel-saving technologies, in high traffic as well as low. Naturally occurring flocks and swarms have long commanded human attention, with much engineering inspiration being drawn from their beauty, order, and cooperation. Recent simulation and modeling of swarms has given rise to interesting mathematical problems as well as useful control strategies for machine applications. To our knowledge, no microscopic, decentralized model of vehicle interactions based on swarming philosophy exists. Here we develop a new model of vehicle interactions on a two-lane highway, made up of ordinary differential equations and smooth functions. The new model’s purpose is not primarily traffic simulation, but cooperative control design. The philosophy behind the modeling is borrowed from the swarm motion control ideas known as ``Reynolds' Rules.'' Vehicles in the swarm have different desired speeds, which can be maintained by changing lanes to avoid slower-moving lead vehicles, while also avoiding both frontal and side collisions. Stability analysis of the proposed model has been presented, as well as simulation results and possible uses. How should a group of cars cooperate in order to save the most fuel? To provide insight into the driving behavioral objectives of a cooperating vehicle group sharing the goal of overall fuel savings, a form of collaborative dynamic eco-driving is developed by applying fuel optimal control to the previously developed swarm-inspired model of vehicle interactions. The new model is used for purposes of control design, where the controller has the objective to minimize the amount of fuel used by the whole group of cars. This is accomplished via optimal control by gradient descent and the method of feasible directions, since the model is nonlinear. The result of this synthesis is a type of collaborative dynamic eco driving that offers insight into the way collaborating vehicles ought to interact on the road in order to save collective fuel.

Comments

© Danielle Fredette, 2017. All rights reserved.

Creative Commons License

Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.

Author Type

Faculty

Link to Full Text

Share

COinS