Adaptive look-ahead model predictive control strategy of vehicular PEMFC thermal management

Abstract

This study presents an improved model predictive control (MPC) strategy for the thermal management of vehicular fuel cells, with the incorporation of model adaptation, the utilization of look-ahead information, and temperature trajectory planning. To describe the thermal dynamics of the fuel cell cooling system, a three-volume model of a 60 kW fuel cell system is proposed and validated with dynamic load experiments. Additionally, a fuel cell temperature sensitivity experiment is conducted to identify a comfort temperature reference. The predictive models developed to cover the entire range of power levels are based on the integration of look-ahead power information and weigh adjustment of actuators. Results from the analysis of look-ahead times suggest that a 4-s preview is suitable for mitigating fuel cell dry and flooding risks. Dynamic control results demonstrate a 22%–60% reduction in temperature tracking error compared to MPC-H, MPC-M, MPC-L and MPID, while energy-saving abilities are increased by 50% relative to MPID, contributing to improve the state balance and economy for vehicular fuel cells.