Neural network approach for the energy management and gear selection of a fuel cell electric vehicle

As the demand for sustainable and low-emission transportation grows, hydrogen fuel cell electric vehicles (FCEVs) stand out as a compelling solution due to their high efficiency, rapid refueling, and zero-emission operation. This paper proposes an advanced energy management strategy for FCEVs based on artificial neural networks (ANNs), aiming to optimize the power distribution between the fuel cell and the battery while enhancing overall system efficiency. The ANN was trained on data generated from a cost function designed to minimize hydrogen consumption, stabilize battery state of charge (SOC), and maximize fuel cell efficiency. A second ANN handled gear selection to enhance motor efficiency. The proposed system was evaluated through real-time hardware-in-the-loop (HIL) simulations under a WLTC driving cycle and compared against a traditional rule-based controller (RBC). Results demonstrate that the ANN-based strategy reduces hydrogen consumption by up to 16.05%, while improving average fuel cell and motor efficiency by 1.75% and 1.13% respectively, confirming its suitability as a robust and scalable solution for next-generation FCEV energy management systems. • Neural networks used for energy distribution and gear selection in fuel cell vehicle. • Controller tested using hardware-in-the-loop under WLTC driving cycle condition. • Hydrogen consumption reduced by 16.05% compared to rule-based control strategy. • Fuel cell efficiency improved by 1.75% and motor efficiency by 1.13% on average.