MPC for battery/fuel cell hybrid vehicles including fuel cell dynamics and battery performance improvement

Abstract

In this paper, the performance and durability of hybrid PEM fuel cell vehicles are investigated. To thatend, a hybrid predictive controller is proposed to improve battery performance and to avoid fuel celland battery degradation. Such controller deals with this complex control problem by handling binaryand continuous variables, piecewise affine models and constraints. Moreover, the control strategy is totrack motor power demand and keep batteries close to a desired battery state of charge which is appro-priately chosen to minimize hydrogen consumption. It is important to highlight the consideration of con-straints which are directly related to the goals of this paper, such as minimum fuel cell power thresholdand time limitation between fuel cell startups and shutdowns. Furthermore, different models have beenelaborated and particularized for a vehicle prototype. These models include few innovations such as a ref-erence governor which smooths fuel cell power demand during sharp power profiles, forcing batteries tosupply such peaks and resulting a longer fuel cell lifetime. Battery thermal dynamics are also taken intoaccount in these models in order to analyze the effect of battery temperature on its degradation. Finally,this paper studies the feasibility of the real implementation, presenting an explicit formulation as a solu-tion to reduce execution time. This explicit controller exhibits the same performance as the hybrid pre-dictive controller does with a reduced computational effort. All the results have been validated in severalsimulations