AI-enhanced energy management strategy of hybrid electric vehicle

Effective, efficient, and real-time energy management strategies are key to reducing fuel consumption in hybrid electric vehicles. However, current energy management strategies suffer from poor real-time performance and fuel economy. At the same time, achieving a balance between real-time performance and optimality in energy management strategies remains a significant challenge. To address these issues, we introduce an energy-saving control framework that comprehensively considers both vehicle speed planning and powertrain energy management. First, we propose a data-driven, high-precision dynamic speed prediction method using neural networks and Markov chains to characterize and analyze the changing patterns of speed disturbance based on realistic vehicle speed data. Furthermore, we utilize reinforcement learning, transfer learning, and other AI methods to achieve real-time adaptive adjustment of vehicular power flow in a complex traffic environment, thus reducing the dependence of traditional energy management methods on accurate vehicle models. Finally, building on the previous studies, we propose a transfer learning-based energy management method for vehicle fleets to improve the efficiency of solving energy management strategies.