High-Precision On-Board Capacity Estimation of Lithium-Ion Cells Using a Fractional-Order Cell Model and Singular Value Decomposition

In order to meet accurate driving range prediction and reliability requirements of Electric-Vehicles (EVs), it is crucial to have an accurate State-of-Health (SOH) estimation of the battery cells. This paper introduces a new on-board capable algorithm for physical model-based estimation of lithium-ion cells' State-of-Health-Capacity (SOHC) based on Differential-Voltage-Analysis (DVA). The algorithm leverages a fractional-order cell model to compensate temperature dependencies of Differential-Voltage (DV) curves and Proper-Orthogonal-Decomposition (POD) to deduce the cell capacity from the DV curves. The algorithm is validated by measurements with a latest EV at temperatures between -20°C and 20°C, confirming overall SOHC estimation Mean-Absolute-Error (MAE) under 1.2%.