AI-Based Power Split Strategy for Hybrid Commercial Vehicle Applications

Abstract

In the rapidly evolving landscape of hybrid commercial vehicle technology, integrating artificial intelligence (AI) with fuel cell applications offers a promising frontier for efficient, sustainable and eco-friendly road freight system. There has been many different approaches on optimization of power split between the electric motor and the fuel cell system (FCS). Conventional approaches use quadratic optimization to determine the optimal power from the electric motor at each discrete grid point along the route, with initial and final battery state of charge (SoC) as constraints. This paper proposes a deep reinforcement learning-based approach to optimize the power split between the electric motor and the FCS in a hybrid vehicle at every time point during the vehicle’s trip. The agent demonstrated the ability to autonomously learn and improve power split decisions, resulting in enhanced fuel efficiency.