Lane-changing trajectory control strategy on fuel consumption in an iterative learning framework

Abstract

A novel lane-changing trajectory control strategy in an iterative learning framework is proposed and its impact on fuel consumption of the off-ramp traffic system is analyzed in this paper. The iterative learning framework includes three layers. In the calibration layer, the lane-changing decision process is modelled by probabilistic neural network while a back-propagation neural network is designed to imitate the lane-changing trajectory. Moreover, we use a cost function and numerical simulation in the optimization layer to optimize the trajectory database and calibrate the proposed framework, respectively. In the application layer, simulation experiments are conducted to examine the fuel consumption of individuals and systems. The results indicate that connected automated vehicles (CAVs) can dissipate quickly after congestion is formed and complete lane-change and avoidance behavior in a more stable state after iterative learning. CAVs can reduce fuel consumption by 35% compared with human-driven vehicles.