Parameter Estimation for a Proton Exchange Membrane Fuel Cell Model Using GRG Technique

Abstract

The parameter estimation of the proton exchange membrane fuel cell (PEMFC) model is important to accurately present the relationship between voltage versus current. Regarding this problem, the difference between observed voltage and model-calibrated voltage, which composed of cell reversible voltage, activation voltage drop, ohmic loss, and concentration voltage drop, should be minimized. So far, various optimization algorithms have tackled this problem. However, there is still a way to improve the solution quality using another technique, and in order to fairly compare the solution qualities among the techniques, more information is required which has been so far missed. Thus, this study proposed generalized reduced gradient (GRG) technique which obtained good results. When compared with two variants of harmony search and two variants of particle swarm optimization, GRG could find much better results in terms of mean square error (MSE). Also, this study provided full problem formulation and numerical dataset, which was scattered in literature and not clearly provided in previous literature. Hopefully, this study invites more researchers to replicate this benchmark problem of the PEMFC parameter estimation and to tackle it using their own techniques in the future.