2021 年 54 巻 4 号 p. 172-183
Accurate state estimation is critical for the management of zinc–nickel single-flow battery (ZNB) stack energy storage systems. The parameters of typically used models are primarily obtained via empirical or offline identification. Hence, the dynamic property of the parameters results in an inaccurate condition monitoring of these models. Therefore, a recursive least squares with forgetting factor algorithm and an unscented Kalman filter algorithm are adopted simultaneously to realize online parameter identification and state estimation. Verification results show that the joint algorithm can accurately capture the dynamic characteristics of the model parameters and exhibits high accuracy and robustness in estimating the capacity and state of charge. To further investigate the potential of ZNBs, the online tracking of peak power is realized by the constant current–constant voltage hybrid mode based on the model above and by considering the design current and cutoff voltage limits. A novel pulse experiment is designed to verify the peak power model, and the results show that the model accuracy satisfies engineering requirements.
