Multi-objective hierarchical energy management for connected plug-in hybrid electric vehicle with cyber–physical interaction

Recently, the overall performances of plug-in hybrid electric vehicles (PHEVs) are expected to be further improved by integrating multiple objectives and cyber–physical interaction (CPI), which pose challenges to existing hierarchical and synchronous optimization frameworks in terms of performances improvement effect and optimization efficiency. In this paper, the CPI based hierarchical optimization framework (HOF) is proposed for the first time to explore the overall performances improvement potential of PHEVs by integrating multi-dimensional traffic information. Particularly, the decomposition based multi-objective evolutionary algorithm is employed in the cyber level to achieve motion planning by balancing various objectives including economy, comfort, safety, and traffic efficiency. Furthermore, the CPI mechanism is developed under the option-critic architecture, which achieves the bi-directional interaction between cyber and physical levels, thus contributing to obtaining superior optimization effect and computational efficiency. Benefiting from the CPI mechanism, the power distribution is finished in the physical level, in which the engine steady-transient characteristic and powertrain efficiency are highlighted comprehensively. The simulation verification with real traffic data demonstrates that the proposed strategy achieves a 4.73% and 5.91% improvement in overall performances compared to synchronous and hierarchical optimization, respectively.