Approximate optimal energy management strategy for multi-speed series-parallel PHEV integrating global prediction and real-time control

The hierarchical energy management strategy (EMS) with a “prediction-control” architecture has been demonstrated to improve the fuel economy of hybrid electric vehicles (HEVs) through the utilization of traffic information. However, the time-varying dynamic characteristics of the driving scenarios result in the accumulation of speed prediction errors, thereby hindering the adaptability of EMS to different driving scenarios. In this paper, a novel approximate optimal EMS for multi-speed series-parallel PHEV integrating global prediction and real-time control is proposed based on the “prediction-control” architecture. First, the global prediction domain predicts the global speed profile to acquire the reference State of Charge (SoC) sequence. Then, the real-time control domain introduces an information evaluation factor (IEF) to construct a multi-objective optimal real-time control framework. Second, a mesoscopic coordination domain that integrates traffic information from multiple scales is established. Within this domain, a high-precision speed prediction method fusing Encoder and BiLSTM is proposed. Subsequently, the IEF is solved online via Gray Wolf optimization combining references from the global prediction domain and feedback from the real-time control domain. This integration provides novel insights for optimizing the control of hybrid system. Finally, results from hardware-in-the-loop demonstrate that the proposed strategy improves fuel economy by nearly 10 % compared to rule-based strategy and shows better adaptability than the hierarchical EMS. This study provides an effective control framework to improve the adaptability of predictive EMS to different driving scenarios.