Energy management of multi-mode plug-in hybrid electric vehicle using multi-agent deep reinforcement learning

The recently emerging multi-mode plug-in hybrid electric vehicle (PHEV) technology is one of the pathways making contributions to decarbonization, and its energy management requires multiple-input multiple-output (MIMO) control. At present, existing methods usually decouple MIMO control into single-output (MISO) can only achieve local optimal performance. To optimize globally, this paper studies a method for PHEV based on multi-agent deep reinforcement learning (MADRL). By introducing relevance ratio, hand-shaking strategy proposed enable two agents work collaboratively under MADRL framework using deterministic policy gradient (DDPG) algorithm. Unified settings DDPG are obtained through sensitivity analysis influencing factors working mode attained parametric study ratio. advantage demonstrated software-in-the-loop testing platform. result indicates that rate greatest factor Using unified ratio 0.2, system save up 4% compared single-agent 23.54% conventional rule-based system.