This paper presents a nonsmooth adaptive extremum seeker that minimizes the hydrogen consumption in a fuel-cell system. The extremum seeker operates by estimating the gradient of the objective function but, unlike other seekers, it does not require a dither signal to produce such estimate. The absence of a dither signal simplifies the choice of parameter values for the seeker, and more importantly, it allows it to converge to the optimal value exactly, not only to a small neighborhood. The proper functioning of the proposed scheme is proved using non-smooth Lyapunov analysis. The strategy is tested on the input–output map of a real polymer electrolyte fuel cells.

Peer Reviewed