Deep Reinforcement Learning Based Fast Charging of Lithium-Ion Batteries with Different Penalty Coefficients

The safe and fast charging is of great significance to improve the practicality of lithium-ion batteries. The rapid temperature rising can increase energy loss and lead to safety problems, which should be taken account in the fast charging optimization process. In this paper, a multi-objective optimal control problem considering charging time, energy loss, and temperature rising is formulated. To solve this problem, a deep reinforcement learning framework incorporated with an electro-thermal model is proposed. The effects of different penalty coefficients on the charging results are presented and discussed. Finally, comparative simulations are conducted to validate the performance of the proposed algorithm. Results show that the proposed charging strategy show better performance than multistage constant charging strategy.