Improved Approach for Online Monitoring of Second Life Lithium-Ion Batteries to Optimize the Performance in Stationary Storage Systems

Monitoring lithium-ion batteries in their second life is of great importance when using them in stationary storage systems for grid support or in any other energy system, due to the expected different behavior and operational conditions in their second life compared to their first life. The idea of this study based on the fact that each estimation method has its advantages and drawbacks depending on the application in which the batteries will be used, and that monitoring batteries in their second life required to be integrated with the battery management system to balance cells, optimize performance, and prevent overheating. The contribution of our study can be summarized in the integration of specific methods, aiming to benefit from most of their advantages and neutralizing their drawbacks as much as possible to provide more accurate and reliable online monitoring of batteries. The combination presented in this study, consisting, for state of charge estimation, of an artificial neural network and an Unscented Kalman Filter with adjusting the filtering process using Coulomb counting, and Kalman Filter for state of health estimation, showed much better reliability and accuracy than if each method was used alone. This combination has been trained and adjusted using empirical data of second-life batteries, extracted from used electric vehicle battery bank, to obtain the best possible performance, considering the specificity of these batteries in terms of operating voltage range, lower current ratio, and other parameters.