Online Estimation of Lithium-Ion Battery State-of-Charge and Capacity with a Multiscale Filtering Technique

Abstract

Real-time prediction of state-of-charge (SOC), state-of-health (SOH) and state-of-life (SOL) plays an essential role in many battery energy storage applications, such as electric vehicle (EV), hybrid electric vehicle (HEV) and smart power grid. However, among these three quantities, only the SOC has been thoroughly studies while there is still lack of rigorous research efforts onthe other two quantities, SOH and SOL. Specially, real- time estimation of the SOH-relevant cell capacity by tracking readily available measurements (e.g., voltage,current and temperature) is still an open problem. Commonly used joint/dual extended Kalman filter(EKF) suffers from the lack of accuracy in the capacity estimation since (i) the cell voltage is the only measurable data for the SOC and capacity estimation and updates and (ii) the capacity is very weakly linked to the cell voltage. Furthermore, although the capacity is a slowly time-varying quantity that indicates cell state-of-health (SOH), the capacity estimation is generally performed on the same time-scale as the quickly time-varying SOC, resulting in high computational complexity. To resolve these difficulties, this paper proposes a multiscale framework with EKF for SOC and capacity estimation. The proposed framework comprises two ideas: (i) a multiscale framework to estimate SOC and capacity that exhibit time-scale separation and (ii) a state projection scheme for accurate and stable capacity estimation. Simulation and experimental results verify the effectiveness of our framework.