Comprehensive Eco Driving Framework for Connected Fuel Cell Hybrid Electric Bus in Highway Based on Improved Soft Actor-Critic Algorithm

Leveraging the communication and intelligent control technology, connected fuel cell hybrid electric vehicles (FCHEVs) allow eco-driving to control power distribution and motion, thereby improving driving performance. This study proposes an improved soft actor-critic (SAC) based eco-driving framework in highway scenario for FCEHVs, co-optimize the energy management system (EMS), longitudinal speed planning, and lateral lane change decisions for driving energy efficiency, safety, comfort, durability, and travel efficiency. Additionally, re-attentive experience replay (RAER) is utilized to flexibly re-evaluate the weights of important experiences to accelerate training. Comparative experiments of different algorithms and scenarios manifest optimality and adaptability of the proposed framework.