Transferred Energy Management Strategies for Hybrid Electric Vehicles Based on Driving Conditions Recognition

TL;DR

This paper introduces a transferred energy management strategy for hybrid electric vehicles that uses driving condition recognition and reinforcement learning to optimize power control, improving fuel efficiency and computational efficiency.

Contribution

It combines driving condition recognition with reinforcement learning to create a transferred EMS that adapts to different driving scenarios in parallel HEVs.

Findings

Enhanced fuel economy demonstrated in simulations

Reduced computational complexity compared to traditional methods

Effective adaptation to various driving conditions

Abstract

Energy management strategies (EMSs) are the most significant components in hybrid electric vehicles (HEVs) because they decide the potential of energy conservation and emission reduction. This work presents a transferred EMS for a parallel HEV via combining the reinforcement learning method and driving conditions recognition. First, the Markov decision process (MDP) and the transition probability matrix are utilized to differentiate the driving conditions. Then, reinforcement learning algorithms are formulated to achieve power split controls, in which Q-tables are tuned by current driving situations. Finally, the proposed transferred framework is estimated and validated in a parallel hybrid topology. Its advantages in computational efficiency and fuel economy are summarized and proved.