State of Charge and State of Health Estimation Using Pseudo Random Binary Sequences and Machine Learning /

Abstract

Batteries are an important category of electrochemical storage technologies that can store electrical energy for remote on-demand consumption. Lithium-Ion Cells are one such technology, that are quickly replacing older batteries like Nickel and Lead Acid Batteries in different applications such as Portable and Smart Devices. They are also being touted as the best option for several future applications such as Hybrid Electric Vehicles, Power Banks, and Alternative Storage, due to their climate friendly nature and several other advantages. However, these cells have a State of Charge and State of Health sensitive operation. These Two key parameters (State of Charge and State of Health) are used to study Cell performance under different loading conditions, and their estimation is the primary focus of this thesis. This thesis proposes an indigenous method of SOC and SOH estimation on a cell level, based on the influence of cell internal impedance on a Pseudo Random Binary Sequence and the derivation of a Piecewise Linear Model through Machine Learning. A hardware model is prepared to verify the method and demonstrate its working. Results of this estimation technique are comparable to other popular PRBS methods but with added simplicity due to less data clustering and quick results.