Multi-model approach for simulation of Li-ion batteries.

The global effort to reduce fossil fuel consumption implies massive energy storage exploitation. The only available technology able to satisfy the required performance is the Li-ion cell. To ensure a sustainable cycle life, it is necessary to exploit the batteries during their overall possible lifetime. BAT-MAN is a project by Brain Technologies to provide a product that offers an online, real-time, and non-intrusive estimation of a lead-acid battery's charge status and state of health. The thesis aims to provide an extension of BAT-MAN to Li-ion batteries. The thesis aims to investigate the behavior of different Li-ion cells to build a general model for simulation and identification purposes. A rigid methodology definition is needed to standardize the experiments and collect consistent data for the successive steps. Some different approaches are investigated to find the best trade-off model between accuracy and computational complexity. Candidate models are tested in closed-loop validation using an Extended Kalman Filter as an observer. The resulting parameters models are examined to extrapolate a simple relation between different SoC and SoH of the batteries. The last phase is the application of the ERMES algorithm, patented by Brain Technologies, to provide a fast and computational inexpensive estimation of the State of Health of the investigated cell.