The optimal design for PEMFC modeling based on Taguchi method and genetic algorithm neural networks

This paper has presented a new approach to estimate the output voltage of proton exchange membrane fuel cell (PEMFC) accurately by combining the use of a genetic algorithm neural networks (GANN) model and the Taguchi method. Using the PEMFC experimental data measured from performance test equipment of PEMFC, the GANN model could be trained and constructed for obtaining the steady state output voltage of PEMFC. Furthermore, in order to determine the important parameters in GANN, the Taguchi method is used for parameter optimization, with the goal of reducing the estimation error. The test equipment of PEMFC is accurate enough for acquiring the output voltage of PEMFC, and is quite useful for teaching purpose. However, taking the high cost, complicated operation procedure and environment safety into consideration, it is necessary to develop a simulation model of PEMFC to benefit teaching and R&D. Therefore, this paper will present an approach for constructing a GANN model with precise accuracy for the output voltage of PEMFC. For achieving the GANN model with high precision, a troublesome work has to be taken care of, that is, to determine all the parameters required in GANN. We will introduce Taguchi method to solve this problem as well. Finally, to show the superiority of proposed model, this approach has compared the estimation values of output voltage for PEMFC from GANN and BPNN models without using Taguchi method. One can easily find that the error of the proposed method is much smaller than that of the GANN model without Taguchi method and of the BPNN model; that is, the proposed approach has better performance on estimation for PEMFC output voltages.

► A new approach is designed by combining the use of a GANN model and the Taguchi method.
► Experiment results of PEMFC are employed to govern the factors’ value for GANN model optimizing.
► Objective of reducing estimation error for PEMFC output voltage is achieved.