Abstract:
Advanced battery management systems (BMS) rely on accurate and computationally efficient models of Li-ion cells to ensure optimal operation, as the performance of BMS
charging and estimation algorithms is directly influenced by the precision of the underlying cell
model. This thesis introduces a state-space based, reduced-order model derived from the Full
Homogenized Macro-scale (FHM) model, an electrochemical model distinct from the commonly
used Pseudo-Two-Dimensional (P2D) model. Unlike the P2D model, which uses volume-averaging
techniques, the FHM model is developed using homogenization theory, enhancing accuracy,
particularly under high temperatures and low states of charge (SoC). To enable real-time
applicability in BMS, the study employs the Discrete Realization Algorithm (DRA) to
approximate the FHM model’s complex, transcendental transfer functions, converting them into
a discrete-time state-space form. This approach preserves model fidelity
while significantly reducing computational requirements. The cell’s current-voltage relationship is
established and validated in this framework work, providing a robust basis for real-time state
estimation. Root mean square error evaluation techniques validate the accuracy of the model, and a
Kalman filter based observer is integrated to further reduce RMSE, enhancing predictive accuracy
for state variables. The resulting model balances computational efficiency with high precision,
making it suitable for BMS applications that demand both speed and accuracy. By providing
insights into electrochemical dynamics, this model is a promising tool for advanced BMS design,
enabling improved safety and performance in Li-ion battery systems for electric vehicles and grid
storage.
