State of Health Estimation of Lithium-Ion Batteries Using Dual Filters and the IMM Strategy

The development of lithium-ion battery (LiB) technology has been driven by society’s collective effort towards greener solutions. LiBs can provide a considerable amount of energy while adding minimal weight to a system. However, these types of batteries can fail unexpectedly due to aggressive current profiles and changing environmental factors. To ensure LiBs operate within their safety levels, a battery monitoring system (BMS) is employed. To ensure reliable functionality of the system, the BMS is required to provide accurate readings of the available state of charge (SOC) in the LiB. This thesis introduces new methods to estimate for the capacity of the LiB under cycling conditions, allowing the BMS to track the state of health (SOH) of the LiB and ensuring accurate readings of SOC. The first contribution of this research involves the implementation of the relatively new sliding innovation filter (SIF) in a dual filter formulation applied to publicly available experimental cycling data. The dual filter was run on a cycle-to-cycle level to estimate for the battery capacity, internal resistance and SOC of the battery. The performance of the SIF was compared with the well-known Kalman filter (KF), and was found to yield more accurate estimates under faulty conditions. The second contribution of this research involves the use of an interacting multiple model (IMM) strategy, formulated with the SIF, to determine the SOH of the battery at different stages of health. The IMM allowed the use of several modes that represented different levels of SOH of the battery. The final contribution of this research is the formulation of a dual-IMM strategy for determining the SOH of a battery under different aging rates. Two publicly available datasets were implemented to test the algorithms under normal and accelerated aging. The dual-IMM strategy demonstrated improved accuracy and robustness when compared with their standard dual filter counterparts.