Predictive energy management for a plug-in hybrid electric vehicle using driving profile segmentation and energy-based analytical SoC planning

This paper presents a new approach to generating reference SoC trajectories for predictive energy management control of plug-in hybrid electric vehicles. Firstly, inspired by an interesting pattern found in globally optimal SoC trajectories, we propose a novel comprehensive procedure to synthesize the reference SoC trajectory design, where intended driving route is divided into multiple segments with different average driving forces and the reference SoC trajectory of each segment is determined using simple analytical formula. Secondly, to facilitate the above planning process, an ordered sample clustering algorithm and a gap statistic algorithm are combined to optimally segment the predicted spatial-domain driving profile data. An adaptive PMP algorithm is finally employed in the lower level to perform instantaneous power split optimization while tracking the planned reference SoC trajectory. Model-in-the-loop test using a high-fidelity forward simulator shows that the proposed approach has superior fuel economy to traditional approach in hilly driving conditions: up to 2.09% fuel saving is achieved. Meanwhile, the proposed approach can obtain near global optimum, with the maximum gap being only 0.49%.