Battery model parameters anomaly corrections with machine learning methods

Lithium-ion batteries play a crucial role in the energy transition, particularly for electric vehicles and renewable energy storage systems. Accurate estimation of battery electrical parameters is essential to ensure effective management of its state of charge (SoC) and state of health (SoH). The 2RC Electric Equivalent Circuit Model is commonly employed, yet its parameter identification is subject to uncertainties that can impair the accuracy of predictions. In this paper, we propose an artificial intelligence-based approach for the detection and correction of anomalies in the estimation of 2RC parameters. Our method relies on the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm for parameter clustering, followed by anomaly detection using the Z -score. Corrections are then applied via a decision tree, which enhances the reliability of the identified parameters. The results demonstrate that the proposed method reduces the number of anomalies in 2RC parameters. A test case scenario showed that 17% of parameters identified by a classical modelling approach were anomalies. The proposed method correct them all leading to a more physical behaviour of model simulations therefore improving model robustness. This approach contributes to enhanced battery management and enhances the performance of energy storage systems.