Estimation and Prediction of Lithium‐Ion Batteries Using a Hybrid Optimized Gaussian Process Regression and Long Short‐Term Memory Transformer

Accurate prediction of the state of health (SOH) and remaining useful life (RUL) of lithium‐ion batteries is critical for ensuring their safe and reliable operation. This article proposes a novel hybrid framework that overcomes the challenges of hyperparameter optimization and long‐term dependency modeling in traditional prediction methods. First, highly relevant health indicators are selected using Pearson correlation coefficients and random forest, ensuring robust inputs for the prediction models. Next, the enhanced whale optimization algorithm optimizes the hyperparameters of the Gaussian process regression model, significantly enhancing its ability to map complex nonlinear relationships in SOH estimation. Finally, a hybrid long short‐term memory and Transformer model is introduced to capture both short‐ and long‐term dependencies within time‐series data for precise RUL prediction. The results demonstrate the superiority of the proposed method, achieving an average mean absolute percentage error of 0.76% for SOH estimation and 0.828% for RUL prediction, thereby improving the reliability and safety of lithium‐ion battery operations.