State estimation for PEM fuel cell systems with time delay by an Unscented Kalman filter and predictor strategy

Polymer electrolyte membrane (PEM) fuel cell systems are investigated as replacement for auxiliary power units (APU) that are currently used for electrical power generation on aircraft. PEM fuel cells are highly efficient energy converters and provide electrical energy, cathode exhaust gas with low oxygen concentration and water. The system investigated is intended for generation of oxygen depleted cathode exhaust air (ODA) on aircraft with ODA-gas having a low oxygen concentration. Measurement of ODA-gas mass flow and oxygen content introduces a significant time delay. For state estimation a reduced order nonlinear model comprising the fuel cell stack, cooling system and cathode exhaust gas dehumidifying system has been derived. In a three steps state estimation strategy involving Unscented Kalman filtering combined with a prediction compensating for the significant time delay the actual system state is estimated. The strategy is implemented in Matlab/Simulink® and has been applied successfully. Experimental results are shown.