Dual-Objective Energy Management for Electric Vehicles Using Digital Twin and Reinforcement Learning
Year: 2026
Type: conference-paper
Abstract: This paper proposes a synergistic utilization of Deep Deterministic Policy Gradient (DDPG) and digital twin technologies in the context of electric vehicles (EVs) that will help achieve an optimal energy management strategy. The intention was first to enhance energy transfer between a Lithium-ion Battery (LIB) and an Ultracapacitor (UC) to counter the depreciation of batteries and increase energy running efficiency. In this regard, a comparative assessment of the Digital Twin Enhanced DDPG, traditional Q-learning and a rule-based method was performed. The findings indicated that Digital Twin Enhanced DDPG had 15% greater energy efficiency and 10% less battery degradation when compared to traditional methods. In addition, the method showed a better State of Charging (SOC) retention indicating the benefit of having real-time feedback provided by the digital twin. As a result, this integration provides a more efficient, sustainable energy management system, which is key when it comes to facilitating electric mobility, as well as demonstrating the potential of digital twin technology in complementing energy management systems based on AI used to manage EVs. Such results provide valuable pointers toward progress in the field.
Language: English
Subfield: Automotive Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7216307229