Real time control of air feed system in a PEM fuel cell by means of an adaptive neural-network

• A neurocontroller based on a B-spline neural network (B-SNN) is proposed.
• The neurocontroller is designed to supply an adequate air flow to a PEM Fuel Cell.
• The B-SNN is based on a simple control algorithm with on-line learning capability.
• Air supply to PEMFC should be strictly controlled during transients of load current.
• B-SNN neurocontroller shows superior performance to control the air supply to PEMFC.

This paper proposes an adaptive B-spline neurocontroller (B-SNN) for the optimal air supply into a Polymer Electrolyte Membrane Fuel Cell (PEMFC). The B-SNN is proposed due to its simple structure, adaptability, and robustness, taking into account the PEMFC's nonlinearities. Furthermore, B-SNN has the capabilities of on-line learning and real time operation. The proposed neurocontroller is tested on a PEMFC hardware-in-loop (HIL) emulator. The FC emulator is based on a detailed nonlinear simulator taken from the open research. Real time results of the PEMFC system emulator and the neurocontroller obtained by the hardware-in-the-loop strategy are exhibited.