Lithium Battery Remaining Useful Life Prediction Based on Attention Mechanism DAE-BiLSTM-Transformer Network

Accurately predicting the remaining useful life (RUL) of lithium batteries is of crucial significance in evaluating the condition of batteries. In this regard, this paper proposes a method. Firstly, the denoising autoencoder (DAE) is used to process the raw data. Noise is mixed with the original input, and after passing through the autoencoder, a lower-dimensional feature representation and reconstructed data are obtained. Then, the reconstructed sequence is input into a two-layer BiLSTM to acquire the characteristics of the data in the time series. Finally, the Transformer is utilized to extract the global and local features. The attention mechanism is introduced, and higher weights are assigned to important features so that the model can focus on the features with greater impact. To verify the proposed method, experiments were conducted on different model combinations using the battery dataset provided by the Center for Advanced Life Cycle Engineering (CALCE) at the University of Maryland. By comparing the results, it was found that the new research model proposed in this paper has better performance. with a MAE of 0.0176 and a RMSE of 0.0273, which verifies the effectiveness and superiority of the research method in this paper.