Joint State of Charge and State of Health Estimation Using Bidirectional LSTM and Bayesian Hyperparameter Optimization

Abstract

In this study, a novel Machine learning-based method for the joint State of Charge and State of Health estimation of Lithium Batteries that tackle real-world applications and with Bayesian Hyperparameter optimization is proposed. The estimated State of Health is used as an input for State of Charge estimation, considering battery degradation. The accuracy and computational cost of the proposed method are compared with the other state-of-the-art Machine Learning models. For the most promising solutions, an in-depth analysis on factors affecting the estimation accuracy is performed. To facilitate further research, a new battery dataset was created using extended dynamic driving cycles, encompassing a wide range of temperature conditions and aging stages. This dataset is publicly available online to support model development and comparative testing by the scientific community. The proposed solution achieves low estimation errors for the whole first life of Lithium Batteries for dynamic applications while providing valuable insights into its applicability and effectiveness in battery energy storage systems.

Keywords

• BiLSTM
• lithium-ion battery
• machine learning
• state of charge
• state of health