In this thesis, a toolkit with the purpose of generating optimal policies fordriving a vehicle with information about upcoming traffic signals has beendeveloped. The toolkit can be used to investigate how to generate the optimalvelocity profile with upcoming traffic signals based on a model of second-by-second fuel consumption. To this purpose, we employ an instantaneous fuelconsumption model and formulate an optimization problem for fuel mini-mization.Following the problem formulation, we explore different numerical waysto solve the minimization problem by discretization. The Runge-Kutta 4 thOrder Method (RK4) is chosen to numerically deal with the differential con-straints in the optimization problem since RK4 gives higher resolution withfewer partitions when discretizing along the time horizon. Then, we turn toDirect Transcription with RK4 Steps and Parallel Shooting (DTRPS) withwhich we translate the minimization problem to a nonlinear programming(NLP) problem.We also include an extensive case study for a door-to-door trip with differ-ent traveling settings: travel during which there are no traffic lights; travelwith one light and travel with two lights. The result shows the capability ofthe toolkit. For a specific setting of a trip, an optimal profile of instantaneousvelocity, acceleration and fuel consumption is generated to achieve the lowestfuel consumption for the entire trip.
【 预 览 】
附件列表
Files
Size
Format
View
On generating driving trajectories in urban traffic to achieve higher fuel efficiency