Detection and Analysis of Anomalies in Timeseries of Li-Ion Battery Field Data

Lithium-Ion batteries are used today in most e-mobility devices. Modelling and understanding these battery systems plays a vital role in being able to predict its behaviour. Identifying anomalies in the battery systems during its operation can help prevent catastrophic failures and increase its safety and viability. Identifying and isolating these anomalous sections from the recorded battery time series also helps to further analyse the anomalous components later. Traditionally, deterministic rule-based state-space approaches are used to model the relationship between the states of a battery to predict its ideal behaviour and compare it with the recorded data to identify the anomalies. In this thesis, the viability of more novel data-driven probabilistic approaches will be explored to model the behaviour of the battery and a framework using these proposed approaches will be used to identify anomalous activity in battery field data.