LAUSR

Online/on-board diagnosis would help to improve fuel cell system durability and output power. Therefore, it is a feature the manufacturers may wish to provide for final users to increase the attractiveness of their product. This add-on requires suitable stack models, parametric identification tools and diagnostic algorithms to be run on low-cost embedded systems, ensuring a good trade-off between accuracy and computation time. In this paper, a computational approach for the impedance parameter identification of polymer electrolyte membrane fuel cell stack is proposed. The method is based on an evolutionary algorithm including sub-population and migration features, which improves the exploration capability of the search space. The goal of the evolutionary algorithm is to find the set of parameters that minimizes an objective function, representing the mismatch between two impedance plots in a normalized plane. The first plot is associated with experimental impedance and the second is computed on the basis of the identified parameters using a circuit model. Three kinds of impedance models, characterized by increasing computational complexity, are used, depending on the experimental data—a linear model made of resistors and capacitors, the Fouquet model and the Dhirde model. Preliminary analysis of the experimental impedance data may evidence correlations among parameters, which can be exploited to reduce the search space of an evolutionary algorithm. The computational approach is validated with literature data in a simulated environment and with experimental data. The results show good accuracy and a computational performance that fits well with the commercial embedded system hardware resources. The implementation of the approach on a low-cost off-the-shelf device achieves small computation times, confirming the suitability of such an approach to online/on-board applications. From a diagnostic perspective, the paper outlines a diagnostic approach based on the identified impedance parameters, on the basis of a small set of experimental data including fuel cell stack faulty conditions.