Robust nonlinear control of fuel cell ultra-capacitor hybrid system

Abstract

In this paper we design robust control strategies for a hybrid solid oxide fuel cell ultra-capacitor system. Fuel cell control is established by using an invariant property of fuel utilization within an input-shaping framework. Two control strategies are developed. The first design uses a nonlinear control approach for which we prove the stability of the closedloop system in presence of system uncertainties. The second uses a standard H ∞ robust control approach. Both strategies enforce the control of State of Charge (SOC) of the ultra-capacitor to a desired level. A hardware-in-the-loop test-stand is developed and experimental results are provided.

Adaptive Control of a Solid Oxide Fuel Cell Ultra-Capacitor Hybrid System

“…5. Details of the simulations and the model can be found in [19] and [29]. As shown, CR allows significantly greater fluctuations in load compared to OL, Fig.…”

“…Hence, we term (2) an invariant relationship of the fuel cell. For a desired steady-state utilization , and a given demanded fuel cell current , a fuel flow demand can be calculated from (2), given by In [19] and [33], it was verified that between D1 and D2, D1 has a more significant impact on transient . To address this, we regulate the fuel cell current by reversing (3) to calculate the regulated current based on the actual fuel flow given by (4) This current regulation approach is illustrated in Fig.…”

“…2, with and defined in (10) and (11), respectively, is uniformly ultimately bounded under the control given in Fig. 6, with and designed as in (10) and (11), respectively, and with parameter adaptation laws given in (13)- (19).…”

“…A strategy for attenuating transient fluctuations in fuel utilization, developed in our prior work [18], [19], will be outlined in this section. For an SOFC system such as in Fig.…”

“…This method was developed in our prior research [18], [19]. For PEM fuel cells, where oxygen starvation is an important issue, existing approaches using filtering, model predictive control (MPC) and robust load governor designs, appear in [7], [20]- [22].…”