Machine learning based Energy management system for better optimisation of power in electric vehicles

An energy management strategy (EMS) is critical for hybrid electric vehicles (HEVs). Future driving circumstances, on the other hand, have a significant impact on the EMS's efficacy. Current EMS techniques mostly adhere to already-existing rules that aren't easily modified. So the EMS may benefit from the environment or the driving cycle itself. Using a DRL-based EMS, this article shows how actions may be selected based on the current state alone, without the need for prediction or previous criteria. An online learning architecture based on DRL is also included in this framework. For HEV energy management, the DRL algorithm is a must-have. MATLAB and the Advanced Vehicle Simulator were used in the co-simulation studies (ADVISOR). The DRL-based EMS is more fuel-efficient than the rule-based EMS. The framework of online learning is very efficient. The suggested approach optimises HEVs in real-time.