remaining useful life

rul prediction

soh estimation

soc estimation

battery management

operating conditions

Journal of energy storage

In order for lithium-ion batteries to function reliably and safely, accurate capacity and remaining useful life (RUL) predictions are essential, but challenging. Some current deep learning-based forecasting methods tend to increase the size of training data and deepen the network structure in an attempt to obtain better predictive results, which is quite resource-intensive. By combining broad learning system (BLS) algorithm and long short-term memory neural network (LSTM NN), a fusion neural network model is developed to outstanding predict the lithium-ion battery capacity and RUL in this work. Specifically, the BLS first produces feature nodes based on the historical capacity data, and applies the enhancement mapping to create enhancement nodes. Afterward, the BLS-LSTM fusion neural network is constructed by concatenating all BLS-created nodes as the input layer of the LSTM NN. Finally, the battery capacity and RUL prediction experiments with different size training sets are conducted to verify the effectiveness of the proposed method based on the battery aging data from the National Aeronautics and Space Administration (NASA) Ames Prognostics Center of Excellence and the Center for Advanced Life Cycle Engineering (CALCE) at the University of Maryland. Experimental results demonstrate that the BLS-LSTM fusion neural network guarantees the precision of the lithium-ion battery capacity and RUL prediction, while the training data can be reduced to only 25% of the whole degraded data. • A novel RUL prediction algorithm combined by LSTM NN and BLS is used to predict the RUL for lithium-ion battery. • The BLS algorithm replaces the input layer of LSTM NN for the purpose of processing the collected signals firstly. • Accurate and fast RUL prediction with only 25% training setting

Lithium-ion battery capacity and remaining useful life prediction using board learning system and long short-term memory neural network