International Journal of Advancements in Electronics and Electrical Engineering
Author(s) : H. A. AALAMI, M. KHORASANI
Proton exchange membrane fuel cells (PEMFCs) can be used in transportation, power plants and distributed generation due to high power density and fast speed of operation. Dynamic modeling of fuel cells is a primary need for performance assessment studies of real-time and controller design. Some new models of Simulink use artificial intelligence for making graph output model. Neural network is one of these models. Neural network just uses input – output data that has obtained in several experiments and does not need to set all the parameters. In this paper, first, the PEMFC has been simulated using feedforward neural networks. This network has been trained using different algorithms and the results were compared to determine the appropriate algorithm by MADM methods. Then the effect of destructive signal on the neural network is evaluated and the authors are attempted to reduce this effect by developing a suitable adaptive filter.