An Overview of Data-Driven Methods for the State of Health Estimation

Nowadays, the widespread adoption of Li-ion batteries in numerous sectors resulted in a significant increase in their usage. This upward trajectory is expected to continue in the foreseeable future. Nonetheless, it is vital to incorporate safety measures to ensure these systems function optimally and reliably. The integration of a sophisticated Battery Management System (BMS) that provides precise estimations of key parameters like the State of Health (SOH) and the State of Charge (SOC) is necessary to prolong the battery's lifetime and enhance its reliability and safety. There are several methods and algorithms to estimate such parameters accurately, and they are divided into model-based and data-driven techniques. Although model-based methods can achieve satisfactory results, their computational cost and model complexity are often too high. On the other hand, the rising development of technology and the plethora of data that are currently available made the black box models to become superior, offering accurate estimations with low computational costs. This paper will present an overview and analysis of the trends in data-driven methods for estimating the SoH of batteries over the past five years.