Reinforcement Learning for Hybrid and Plug-In Hybrid Electric Vehicle Energy Management: Recent Advances and Prospects

Energy management is a critical technology in plug-in hybrid-electric vehicles (PHEVs) for maximizing efficiency, fuel economy, and range, as well as reducing pollutant emissions. At the same time, deep reinforcement learning (DRL) has become an effective and important methodology to formulate model-free and realtime energy-management strategies for HEVs and PHEVs. In this article, we describe the energy-management issues of HEVs/PHEVs and summarize a variety of potential DRL applications for onboard energy management. In addition to the control objective and constraints, an elaborate model of the powertrain components is necessary as part of the solution. For example, the modeling of an engine involves the calculation of the fuel consumption, an estimate of efficiency, and the derivation of the torque and angular speed. Also, a computation of efficiency and an expression of the power balance are required for motor modeling. The transfer process of the speed and power from the motor/generator to the final drive is part of the transmission modeling.