Data-Driven Intelligence for Battery Diagnostics and Prognostics in Energy Storage Systems

Lithium-ion batteries are central to electric mobility and grid-scale energy storage, yet their safe and efficient operation depends on accurate estimation of hidden internal states and timely prediction of degradation. This seminar presents a data-driven framework for battery diagnostics and prognostics using machine learning models developed across cell- to pack-level applications. The talk will cover diagnosis through state estimation, including simultaneous state-of-charge and state-of-energy prediction etc., and prognosis through time-ahead capacity degradation modeling. Experimental validation using open-source and in-house datasets, along with real-time battery platform testing, demonstrates practical feasibility for battery management systems. The framework enables scalable health monitoring, enhances reliability, and supports predictive analytics for energy storage systems under real-world operating conditions. The presented work concludes with providing some future directions in scalable health monitoring, degradation tracking, and predictive battery analytics for advanced energy storage systems.