Thermal fault-detection in series connected Li-ion cells: a set-based approach using constrained zonotopes.

Advanced Battery Management Systems (BMSs) rely on mathematical models to enhance battery safety and improve reliability. Operating conditions and temperature have an impact on the performance, longevity and safety of Li-ion batteries. In particular, thermal breakdowns are caused by faults of different physical and chemical nature. In this paper, we use a set-based fault detection approach with unknown but bounded uncertainties to address the challenge of detecting thermal faults of Li-ion series-connected cells. The proposed approach relies on Constrained Zonotopes (CZ), a set representation capable to preserve the computational benefits of zonotopes while describing asymmetric convex polytopes. An equivalent circuit model (ECM) together with a lumped thermal model are used for state estimation taking into account bounded parametric uncertainties and measurement noise. Numerical simulations demonstrate the effectiveness of the proposed strategy in detecting thermal faults at an early stage and indicate the benefits. The results corroborate the advantages of constrained zonotopes over zonotopes.