Battery longevity-conscious energy management predictive control strategy optimized by using deep reinforcement learning algorithm for a fuel cell hybrid electric vehicle

Abstract

Energy management strategies play an essential role in improving fuel economy and extending battery lifetime for fuel cell hybrid electric vehicles. However, the traditional energy management strategy ignores the lifetime of the battery for good fuel economy. To overcome this drawback, a battery longevity-conscious energy management predictive control strategy is proposed based on the deep reinforcement learning algorithm predictive equivalent consumption minimization strategy (DRL-PECMS) in this study. To begin with, the back-propagation neural network is devised for predicting demand power, and the predictive equivalent consumption minimum strategy (PECMS) is proposed to improve the hydrogen consumption. Then, in order to improve the battery durability performance, the deep reinforcement learning algorithm is utilized to optimize the battery power and improve battery lifetime. Finally, numerical verification and hard-ware in the loop experiments are conducted to validate hydrogen consumption and battery durability performance of the proposed strategy. The validation results show that, compared with CD/CS and SQP(Sequential Quadratic Programming), the PECMS combined can achieve better fuel economy with the fuel consumption reduction by 55.6 % and 5.27 %, which effectively improves the fuel economy. The DRL-PECMS can reduce the effective Ah-throughput by 3.1 % compared with the PECMS. The numerous validations and comparisons demonstrate that the proposed strategy effectively accomplishes the trade-off optimization between energy consumption and battery durability performance.