Remaining useful life prediction of lithium-ion battery based on discrete wavelet transform and GRU neural network

With the promotion of energy conservation and emission reduction in China, lithium-ion batteries as a clean energy source have been widely used in military, aerospace and other fields due to their own characteristics. However, due to the possible hazards of lithium batteries, a reasonable prediction of lithium battery RUL is necessary. Traditional prediction methods have lower prediction accuracy and stability due to the capacity regeneration problem existing in lithium batteries. In this paper, the discrete wavelet algorithm is used to decompose and reconstruct the data. After better studying the data features, the battery RUL is predicted using the GRU neural network. Through the lithium battery data slet provided by NASA, the prediction results of the DWT-GRU model are compared with the GRU model, and the results show that the prediction accuracy is improved by nearly 30%, which proves the accuracy and robustness of the method.