State of Charge Estimation of Single-Flow Zinc-Nickel Batteries Based on the Improved Unscented Kalman Filter and Extended Kalman Filter Algorithm.

Single-flow zinc–nickel batteries are a novel type of flow batteries that feature a simple structure, large-scale energy storage capacity, and low cost. The state of charge (SOC) is a crucial indicator of battery performance, providing essential data for the management and control of the battery management system. However, in the estimation of SOC using a traditional unscented Kalman filter, the covariance matrix P often falls into a non-positive definite state during the decomposition steps, leading to algorithm failure. To address this issue, this paper incorporates the singular-value decomposition method into the unscented Kalman filter, resulting in an improved unscented Kalman filter algorithm. Considering the potential low accuracy of a single filtering method for SOC estimation, the improved unscented Kalman filter algorithm is combined with the extended Kalman filter for joint estimation. Experimental comparisons demonstrate that the improved unscented Kalman filter and extended Kalman filter joint estimation achieve higher estimation accuracy compared to using the improved unscented Kalman filter alone. [ABSTRACT FROM AUTHOR]