A novel state energy spatialization regenerative braking control strategy based on Q- learning algorithm for a super-mild hybrid electric vehicle

ABSTRACT

Most of the existing regenerative braking control strategies maintain the battery balance by the state of charge (SOC) penalty function. However, the SOC penalty function only realizes the instantaneous balance and ignores the global balance. To solve this problem, this paper proposes a novel state energy spatialization regenerative braking control strategy. Firstly, the battery energy is converted from the time domain to the space domain at each state. Based on the principle of balance between driving energy consumption and braking recovery energy, the mathematical model of referenced recovery energy is established. Then, considering the maximum energy recovery and the global battery energy balance, the return function is formulated, which takes the referenced recovery energy as the constraint condition. Subsequently, the proposed strategy is optimized by Q-learning algorithm and the optimal motor torques of regenerative braking are obtained. Finally, the Kullback–Leibler (KL) divergence rate is adopted to recognize the type of actual driving cycle, and the online optimal motor torque is obtained by looking up the corresponding table. Using the MATLAB/Simulink software, the simulation model of real working condition in Yubei district of Chongqing is established. The simulation results show that the SOC variation of the proposed strategy is 33.5% and 20.1% lower than that of the maximum energy recovery and DP strategy, respectively. The results indicate the proposed strategy maintains the global balance of battery energy better than the conventional strategy.

KEYWORDS:

• State energy spatialization
• Q-learning
• Kullback–Leibler divergence rate
• regenerative braking
• HEV

Data availability

The data used to support the findings of this study are included within the article.

Acronyms, parameters, and variables used across the paper

Table

Table

Additional information

Funding

This research is supported by the Scientific and Technology Research Program of Chongqing Municipal Education Commission (Grant No. KJQN201800718) and Natural Science Foundation Project of Chongqing (Grant No. cstc2016jcyjA0467); Natural Science Foundation Project of Chong qing [cstc2016jcyjA0467].