Review of adaptive energy management considering driving style and traffic state for new energy vehicles

Reviewartikel, 2026

With the large-scale deployment of new energy vehicles (NEVs) and increasingly complex operating environments, energy management systems (EMS) are challenged by heterogeneous driving behaviors and dynamically varying traffic conditions. Driving-style variability reshapes energy consumption patterns, while time-varying traffic disturbances demand timely, context-aware control, making static and single-variable strategies inadequate for personalized and responsive energy management. Accordingly, this review systematically examines two key control features, namely driving style and traffic state, and clarifies their roles in the modeling, optimization, and implementation of EMS strategies. First, dynamic control based on driving style recognition is reviewed. Feature modeling and identification methods are summarized, and three categories of control mapping mechanisms are synthesized, including parameter adjustment, policy switching, and integrated optimization. Strategy evolution techniques that enable adaptive control are further discussed, such as reinforcement learning (RL) and digital twin (DT) technologies. Second, real-time optimization driven by traffic prediction is analyzed. Perception modeling and multi-horizon forecasting methods are reviewed, their roles in feedforward control and multi-objective scheduling are assessed, and the integration of techniques such as machine learning (ML) within the perception–prediction–control loop is described. Furthermore, unified control architectures that jointly incorporate driving style and environment states are explored, emphasizing co-design for heterogeneous perception, joint decision-making, and system deployment. Finally, current challenges are identified in model integration depth, policy transferability, and system-level feasibility, and future directions are outlined in semantic modeling, graph-based coordination, behavioral evolution, and federated optimization to support the intelligent advancement of EMS.