Fast Battery SoC Trajectory Planning for Predictive Energy Management of PHEBs

Abstract

The battery state of charge (SoC) trajectory is an essential characteristic that represents the balance of consuming fossil fuel and electric energy in hybrid electric vehicles. An optimally planned global SoC trajectory is extremely helpful to the energy management of hybrid powertrains. In this paper, a fast SoC planning method based on supervised learning is proposed, while the global driving cycle is forecasted. The planned SoC trajectory is applied as a guidance for model predictive control (MPC) of the hybrid powertrain in real-time. Real driving cycles are collected from a plug-in hybrid electric bus (PHEB) running in Zhengzhou, China. The SoC planning accuracy and its performance in improving the vehicle fuel economy are validated through a comparison with dynamic programming (DP) results. Simulations demonstrate that our proposed fast SoC planning method is able to reduce the computation time from several minutes to within 1 seconds, and the fuel economy improvement is over 40%.