A Novel Battery RUL Prediction Approach Based on Probabilistic Hyperparameter Optimization

The remaining useful life (RUL) of lithium-ion batteries, a pivotal metric for assessing battery performance and service life, is not amenable to direct measurement. To refine the precision of estimating the remaining useful lifespan for battery systems, this paper proposes a prediction method based on the Bayesian optimization algorithm for Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) neural networks. The research utilized the battery dataset provided by the MIT Stanford Toyota Research Institute. The method first preprocesses the battery data, selecting key health indicators (HI) through the Pearson correlation coefficient, and then estimates the RUL of the battery using LSTM and GRU networks. In order to augment the efficacy of the RUL estimation model, this paper applies Bayesian optimization technology to determine the optimal hyperparameters, thereby saving time required for model prediction.