Model Predictive Control with Gaussian Process Regression Compensation for Power Management in Fuel Cell Hybrid Electric Buses

Abstract

Fuel cell hybrid electric buses (FCHEBs) using hydrogen fuel cells (FCs) as the main power source and supercapacitors (SCs) as the energy buffer may be a viable electrified transportation technology. This paper proposes a Model Predictive Control scheme with Gaussian Process Regression Compensation (GPRC-MPC) for power management in FCHEB FC/SC hybrid energy storage systems. To improve the accuracy of the linear MPC model, GPRC-MPC introduces Gaussian process regression to predict and compensate for the load disturbance error and the residual error. Simulation results show that GPRC-MPC performs better in reducing the hydrogen consumption, maintaining the SC SOC level, and alleviating the FC degradation.