Neural Network Approach for Semi-Empirical Modelling of PEM Fuel-Cell

• Published in: 2006 IEEE International Symposium on Industrial Electronics Vol. 3; pp. 1858 - 1863
• Main Authors: Hatti, M., Tioursi, M., Nouibat, W.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.07.2006
• Subjects: Biomembranes; Design optimization; Fuel cells; Neural networks; Power generation; Power system modeling; Power system simulation; Protons; Temperature distribution; Voltage
• ISBN: 9781424404964; 1424404967
• ISSN: 2163-5137
• DOI: 10.1109/ISIE.2006.295855

In this paper we consider proton exchange membrane fuel cells due to low working temperature (80-100degC), fast start up and a relatively simple design make PEMFCs strong candidates to provide power plants suited for residential, vehicular applications and for broad range of systems, including the next generation of non-polluting automobiles, distributed power generation, and portable electronic appliances. The simulation of PEMFC can work as a powerful tool in the development and widespread testing of alternative clean and environmentally acceptable energy source, To improve the system performance, design optimization and analysis of fuel cell systems are important. Mathematical models and simulation are needed as tools for design optimization of fuel cells, stacks, and fuel cell power systems. The aim of our work was to develop a model that includes all important operating characteristics of the processes using non-parametric approach. Here, the Levenberg-Marquardt neural network method was employed in the modeling of PEMFC, and has shown good performance in the prediction of cell voltages. Specifically, we have been able to model the effect of stoichiometric parameters on the cell voltage. The trained ANN model is computationally fast and easy to use, especially in the cases where physical models are not readily available