Velocity prediction and profile optimization based real-time energy management strategy for Plug-in hybrid electric buses

The Plug-in hybrid vehicle (PHEV) has been progressively penetrated in the urban public transport system and seen a foreseeable fast growth in the future. Within this horizon, energy management is an enabling technique for the cost-efficient operation of the PHEV. In this paper, a model predictive control (MPC)-based real-time energy management strategy (EMS) combining a cloud-enabled velocity profile optimizer (VPO) and vehicle-side velocity predictor is proposed for the Plug-in hybrid bus (PHEB) under the intelligent transportation systems (ITS). Particularly, the velocity profile and the state of charge (SOC) sequences are optimized by incorporating the genetic algorithm (GA) with the dynamic programming (DP), giving rise to a novel GA-DP-based VPO. In the case that the vehicle can be hardly decoupled from the traffic flow, a multi-feature predictor based on Long Short Term Memory (LSTM) Network is triggered to replace the cloud-enabled VPO to predict the short-term velocity. Results show that the prediction accuracy can be improved by 5.4% by employing the multi-feature training. The equivalent fuel consumption with the mode-switching EMS in the optimized UDDS cycle can be reduced by 14.9% compared with the state of the art. The proposed strategy is validated with a real-time performance by performing the hardware in the loop (HIL) experiment.