State Estimation of a Molten Carbonate Fuel Cell by an Extended Kalman Filter

Abstract

fuel cell stacks only provide very limited measurement information. To overcome this deficit, a state estimator for a molten carbonate fuel cell is developed in this contribution. The starting point of the work is a rigorous spatially distributed model of the system. Due to its complexity, this model is hardly suitable for the design of a state estimator. Therefore, a reduced model is derived by using a Galerkin method and the Karhunen Loève decomposition technique. A low order system of ordinary differential equations and algebraic equations results. The reduced model is used to study the observability of the system for different sensor configurations. An extended Kalman filter with a continuous time simulator part and a discrete time corrector part is designed on the basis of the reduced model. The filter is tested in simulations.

Keywords: Fuel cell, MCFC, partial differential equations, model reduction, Karhunen Loève decomposition, state estimation, Kalman filter.