A Hybrid Exponential Smoothing and LSTM Approach With Double Lightning Search for Enhanced Prediction of Lithium‐Ion Battery Remaining Useful Life

ABSTRACT Lithium‐ion batteries are essential in modern energy systems, supplying electricity to many consumer products that require dependable operation. This study introduces a novel method for predicting the remaining useful life (RUL) and end‐of‐life (EoL) of lithium‐ion batteries (LIBs) using the long short‐term memory (LSTM) model. The process includes getting the data ready, using exponential smoothing (ES), making a sequence, using the double lightning search algorithm (DLSA) to find the best hyperparameters, and doing iterative LSTM modeling. ES is utilized to preprocess data, reduce noise, and improve input quality for the subsequent model phases, while the iterative integration of LSTM modeling and DLSA hyperparameter optimization enhances capacity estimates across the entire battery lifecycle. The model performance is assessed using basic statistical measures, indicating consistent accuracy in predicting the state of health (SOH) and RUL. The way the time series is set up lets the LSTM accurately find important time‐based relationships, even when there is much noise. Adaptive hyperparameter selection through DLSA safeguards against fluctuating degradation rates, yielding an average error of under ±0.002 ampere‐hours in capacity prediction. Systematic sampling and normalization make calculations faster without changing the quality of the data. Data from the NASA Prognostics Center of Excellence on LIB performance and degradation are used to validate the model. This predictive modeling approach shows how data smoothing, deep learning, and intelligent search algorithms can improve RUL and EoL forecasts for real‐time battery health monitoring and operational safety.