Moving horizon state estimation for a PEM fuel cell system based on successive linearisation

Abstract

Due to the growing demand for sustainable and environmentally friendly energy sources,the fuel cell has become the focus of science once again. If the fuel cell is operated with green hydrogen, it represents an entirely CO2-neutral alternative to conventional energy generation methods. A major challenge in the operation of the fuel cell is that it is sensitive to certain conditions and disturbances. In order to ensure efficient and reliable operation of a polymer electrolyte membrane fuel cell (PEMFC), the condition of the fuel cell has to be monitored. Since the relevant internal states cannot be measured, they can be reconstructed, however, with the help of state observers. This thesis deals with developing and implementing a moving horizon estimator (MHE) for a PEM fuel cell model. First, a zero-dimensional model of the fuel cell is developed, which contains the main effects and represents the basic system behaviour. Subsequently, a formulation of the MHE is derived based on successive linearisation, which enables an implementation of the observer concept with common quadratic solvers. The validation of the state estimation method is carried out on the PEM fuel cell model and the 3-tank model,which are both nonlinear systems. Finally, various properties of the MHE are analysed in more detail, and the observer is compared with an extended Kalman filter (EKF). The simulation results show that the MHE can observe the states with satisfactory accuracy based on successive linearisation on both systems. The investigations further show that the derived formulation of the observer is sufficiently accurate and that the MHE can be used to reconstruct the initial state in the case of a poor initial estimation as well. The derived MHE convinces with its convergence properties and the beneficial matrix based formulation.