Progress in prediction of remaining useful life of hydrogen fuel cells based on deep learning

Abstract

Hydrogen fuel cells are promising power sources that directly transform the chemical energy produced by the chemical reaction of hydrogen and oxygen into electrical energy. However, the life of fuel cells is the main factor restricting their large-scale commercialization; therefore, it is crucial to predict their remaining useful life (RUL). In recent years, deep learning methods for RUL prediction has shown promising research prospects. Deep learning methods can improve the accuracy and robustness of predictions. In this study, the RUL prediction of hydrogen fuel cells based on deep learning methods was systematically reviewed, and various methods were compared. First, the characteristics and applications of different types of fuel cells were reviewed, and the benefits and drawbacks of three RUL prediction methods were compared. Second, different deep learning methods used to predict fuel cell RUL, such as convolutional neural networks (CNN), recurrent neural networks (RNN), Transformer, other algorithms, and fusion algorithms, were systematically reviewed, and the performance and characteristics of different algorithms were analyzed. Finally, the aforementioned research was discussed, and future development trends were prospected.