Published Paper Details:

Hybrid electric vehicles (HEVs) have the potential to reduce fuel consumption and emissions while improving the performance of conventional vehicles. However, the design of an energy management strategy for HEVs is a challenging problem due to the complex interactions between the vehicle's power sources and the driving conditions. The study commences with a comprehensive understanding of the problem domain, emphasizing the need to balance the power allocation between the internal combustion engine (ICE) and the electric motor to meet performance requirements while minimizing fuel consumption. Data collected from various vehicle sensors are preprocessed and used to train a DRL agent, implemented through neural network architectures. In this paper, we propose a new energy management strategy for HEVs based on DRL. The proposed strategy uses a DRL agent to learn how to control the power flow between the HEV's power sources in order to minimize fuel consumption and maximize range. The DRL agent is trained using a dataset of driving cycles that are representative of the operating conditions of the HEV. The performance of the proposed strategy is evaluated on a variety of driving cycles and it is shown to be able to significantly improve the fuel economy of the HEV. The proposed strategy has several advantages over traditional energy management strategies. First, it is able to adapt to changes in the driving conditions. Second, it is able to learn the optimal control policy even in the presence of uncertainty. Third, it is able to handle complex interactions between the vehicle's power sources. The proposed strategy is a promising approach for developing energy management strategies for HEVs. The results of this study show that DRL can be used to develop efficient and robust energy management strategies for HEVs.