Remaining Useful Life of Lithium-Ion Battery Using Hybrid Long-Short Term Memory and Modified Gravitational Search Algorithm

The rapid development of electric vehicles (EVs) plays a key role in achieving net-zero emissions by shifting energy sources from conventional fossil fuels to renewable energy source.This study introduces a novel hybrid algorithm that integrates long short-term memory (LSTM) with a modified gravitational search algorithm (MGSA) called MGSA-LSTM to enhance the prediction of remaining useful life (RuL) of lithium-ion batteries (LIBs).The MSGA is employed to optimize the LSTM parameters including batch size, number of hidden neurons, and learning rate.The improvement in MGSA involves replacing the velocity formulation in GSA with the velocity model from particle swarm optimization (PSO) to improve global search efficiency, as the GSA tends to find the best value based on the local optima.Evaluated on battery datasets from NASA Prognostics Center of Excellence Data Set Repository (B0005, B0006, B0018), the proposed method achieves root mean square error (RMSE) values of 0.00930, 0.01188, and 0.20972, respectively.These results indicate that this proposed approach offers a promising technique with the potential to significantly improve predictive accuracy in battery management systems for electric vehicles.