Lithium-ion batteries: fast simulation, set-based state estimation and fault detection

Lithium-ion batteries have become in the last years a fundamental component for electrical energy storage thanks to the growing efforts in reducing fossil fuel consumption. Typical applications are portable electronics and electric/hybrid vehicles. The management of lithium-ion batteries is a challenging task handled by an appropriate device called Battery Management System (BMS). State estimation of fundamental quantities, such as State of Charge, is among the routines of the BMS. A direct application of state estimation is fault detection, where the estimated states are compared to the measurements obtained. If a discrepancy is noted, a fault might have occurred. In this work, a set-based method is proposed, based on constrained zonotopes. They are able to compute accurate enclosures and yield better results in state estimation than other set-based methods (e.g. intervals). In addition, if the battery pack is large, computations can become expensive. This works aims at solving this issue by proposing an acceleration method based on Waveform Relaxation techniques. The system of equations representing the dynamics is split in smaller subsystems that can be solved faster. Convergence to the same result of the original problem is obtained by iterating the resolution of the subsystems and applying appropriate convergence conditions.