An Estimation Method for State of Charge of Lithium-ion Batteries Based on Adaptive Multi-Innovation Fractional Order Unscented Kalman Filtering

Accurate state of charge (SOC) is the basis of battery management system for lithium-ion batteries. In this paper, an adaptive multi-innovation fractional order unscented Kalman filtering (AMIFOUKF) is proposed to estimate SOC. Firstly, a fractional order model is constructed, which are identified by adaptive genetic algorithm. Secondly, to address the estimation bias caused by the fixed noise and the single innovation in unscented Kalman filtering, an AMIFOUKF is proposed by adaptive algorithm and multi-innovation identification theory. The proposed AMIFOUKF is verified by the datasets under urban dynamometer driving schedule condition. The experimental results demonstrate that the proposed AMIFOUKF can achieve high accuracy of SOC estimation, the error of estimated SOC and predicted voltage are less than 0.58% and 0.52%, respectively.