SoC Planner for Predictive Energy Management of Fuel Cell Vehicles

Abstract

Global state of charge (SoC) trajectory planning is of great significance for the fuel economy improvement of plug in fuel cell vehicles (PFCV), which is equipped with a relatively larger battery pack. In this paper, we provide a comprehensive analysis of three SoC planners, applied within a model predictive control (MPC) framework. The basic principles of rule-based, dynamic programing based and neural network based SoC planners are described systematically. Their planning accuracy and computation efficiency are compared. The generated reference SoC profile is used for global guidance of real-time MPC energy management. The fuel performance of guided MPC under different reference SoC is further analyzed. The simulation results demonstrate that the fuel economy of real-time algorithm can be significantly improved by introducing a reference SoC (3.31∼4.55%), while the impact of reference SoC precision is not obvious (0.25∼1.24%).