Remaining useful life prediction of PEMFC based on informer model

The prediction of the remaining useful life (RUL) of proton exchange membrane fuel cells (PEMFC) is one of the reasons limiting their large-scale application, and is a key focus of current researchers. Time series prediction methods for PEMFCs based on sliding windows, such as CNN and LSTM, have larger errors as the longer the time series prediction. To address the issue of error accumulation in the RUL prediction of PEMFC using traditional time series methods, this paper innovatively introduces the Informer model into the RUL prediction of PEMFC. Informer builds upon the Transformer architecture by incorporating a self-attention mechanism to mitigate the accumulation of prediction errors and improve the prediction effect of the model. By comparing the predictive results of the static and semi-dynamic PEMFC datasets with those of the traditional LSTM, it is verified that the Informer model has excellent performance in PEMFC time series prediction. The predictive results indicate that the Informer can significantly improve the accuracy of the RUL prediction of PEMFC to a certain extent.