关键词: 质子交换膜燃料电池, 剩余使用寿命预测, 哈里斯鹰算法, 长短期记忆网络, 自注意力机制

Abstract: Proton Exchange Membrane Fuel Cells (PEMFC) have broad application prospects in many fields. Accurate prediction of their Remaining Useful Life (RUL) can monitor the internal aging situation of the batteries in real time and reduce the usage risks of the batteries. Based on the changing trend of the power of PEMFC over time, this study proposes an improved RUL prediction model for PEMFC that combines the Harris Hawks Optimization (HHO), Long-Short Term Memory (LSTM), and Self-Attention Mechanism (Self-Attention). Firstly, the time-power variation curve is obtained based on the relationship between current and voltage data. The combination of Wavelet Absolute Deviation (WAD) and Empirical Mode Decomposition with Ensemble Empirical Mode Decomposition (ES) is used to decompose, denoise and reconstruct the time-power data. Then, in view of the shortcomings of LSTM such as excessive training parameters and large computational load, a method that combines Logistics chaotic mapping and HHO to optimize LSTM is proposed to improve the training speed and prediction accuracy of the model. Finally, based on the advantages of Self-Attention in focusing on key information and improving the accuracy of model training, the HHO-LSTM-Self-Attention prediction model is constructed. Comparative experiments with existing models are carried out, and the results show that the proposed model has higher prediction accuracy.

Key words: Proton exchange membrane fuel cell, Remaining Useful Life prediction, Harris eagle algorithm, Long-short memory network, Self- attention mechanism