Capacity Degradation Modeling for Li-Ion Batteries using a Multiscale Gamma Process Approach

Prediction and analysis of dynamic degradation systems have become an important aspect in the field of prognostics and health management (PHM). Complex applications, such as monitoring the health of a battery system, requires substantial analysis with hybrid methodologies. In recent studies, data-driven methods combined with filtering techniques have shown satisfactory performance [1]. Successful combinations include machine learning approaches, such as neural networks, and nonlinear filtering algorithms, such as extended Kalman filter (EKF). Specifically, the filtering procedure is utilized to concurrently estimate both the state of interest and the parameters of the data-driven model.