Hydrogen consumption estimation of fuel cell vehicle based on vehicle energy transfer

The hydrogen consumption of fuel cell vehicles (FCVs) can serve as both a measure of vehicle economy and a real-time notification of the driver’s fuel surplus status while driving. The traditional estimation method has poor real-time performance and a large error in accuracy. This paper proposes a new method for predicting hydrogen consumption based on vehicle energy transfer. This method extracts relative parameters based on the models of various components during the hydrogen conversion process and then trains the hydrogen consumption prediction model using the Informer algorithm. The training results show that the model has high prediction accuracy. Finally, the accuracy of this method was compared with the LSTM algorithm under urban and high-speed conditions. The results show that the root mean square error (RMSE) and mean absolute error (MAE) are optimized by 35 % and 26 % under urban conditions, while the RMSE and MAE are optimized by 41 % and 36 % under high-speed conditions. Its maximum error remains below 4 g.