1. Optimal sizing and learning-based energy management strategy of NCR/LTO hybrid battery system for electric taxis.
- Author
-
Niu, Junyan, Zhuang, Weichao, Ye, Jianwei, Song, Ziyou, Yin, Guodong, and Zhang, Yuanjian
- Subjects
- *
ENERGY management , *ELECTRIC batteries , *REINFORCEMENT learning , *REAL-time control , *HYBRID electric vehicles , *TAXICABS - Abstract
This paper proposes an offline sizing method and an online energy management strategy for the electric vehicle with semi-active hybrid battery system (HBS). The semi-active HBS is composed by Nickel Cobalt Rechargeable (NCR) and lithium titanate (LTO) batteries with a bi-directional DC/DC converter. First, the vehicle dynamics and the HBS are modelled. Second, a hierarchical optimal sizing method is proposed to minimize the distance-based cost (DBC) of electric taxi in a variety of driving cycles. The lower layer optimizes the energy management strategy (EMS) with dynamic programming (DP), while the upper layer optimizes the sizes of HBS for minimum DBC. Based on the sizing results, the DBC decreases firstly and then increases with the increasing LTO size. In addition, the results of DP indicate the SOC of the LTO batteries works between 50% and 80% for optimal NCR lifespan. Third, by using the rule extracted from DP, a learning-based EMS, i.e., deep deterministic policy gradient (DDPG), is proposed with excellent real-time control potential. Finally, the simulation results show that the proposed DDPG EMS achieves the improved performance than fuzzy logic control EMS and closed result with what can be achieved through DP, yet the computation time is much less. • NCR/LTO hybrid battery system is proposed. • A hierarchical framework is proposed to optimize the size of HBS. • Rule-based strategy is extracted from the optimal EMS. • DDPG is proposed to achieve the near-optimal real-time control. [ABSTRACT FROM AUTHOR]
- Published
- 2022
- Full Text
- View/download PDF