Modeling an ANN-based control for optimal operation of PEMFC systems

In the field of the proton exchange membrane fuel cells (PEMFCs), the main objectives of a control strategy are to optimize the electrochemical reactions in the reaction subsystem in order to increase the fuel economy and energy efficiency, as well as ensure the load demand under a dynamic load profile and variable available power from renewable energy sources (RESs). Among others, the Global Extremum Seeking (GES) optimization technique has shown its capability to deal with these requirements. In this paper a classical implementation of a GES controller is analyzed and an artificial neural network (ANN) based implementation is obtained, showing that the approximation is accurate.