A Data-Based Prognostic Technique of a Proton Exchange Membrane Fuel Cell Applying Dynamic Load Cycle

The present paper proposes a new method for predicting the degradation of the proton exchange membrane fuel cells (PEMFC). The prognostic technique is composed of two different predictors and combines its results to have a more accurate prediction. The first predictor forecast the long-term trend of the degradation in the voltage. The second predictor uses a neural network to improve the accuracy of the prognostic. The proposed prognostic is implemented in a dataset with a dynamic load cycle. The combined prediction has an accuracy upper than 88.53% for the correlation coefficient and a root mean square error (RMSE) lower than 0.0042 with a prognostic horizon of 600 hours.