Hardware-in-the-Loop Experimental Validation of a Learning based Neuro-Fuzzy Energy Management Strategy for Plug-in Hybrid Electric Buses

Abstract

Learning based energy management strategies are promising methods, due to the design complexity minimization and learning capability from historical data. This paper aims to experimentally validate a learning based neuro-fuzzy energy management strategy for plug-in hybrid electric buses. With the aim to minimize computation cost and further improve the energy management strategy, this energy management strategy is designed based on the dynamic programming optimal operation of different auxiliary consumption levels, designed based on the neuro-fuzzy learning technique. The developed learning based neuro-fuzzy energy management strategy has been implemented into a physical control hardware and validated in real-time, managing the energetic operation of an emulated plug-in hybrid electric bus. Fuel consumption decrease of 10.32% has been achieved compared to a charge-depleting charge-sustaining energy management strategy.